CDC Guide For Linux, Unix, and Windows: Informatica Powerexchange (Version 9.0)

Informatica PowerExchange (Version 9.
0)
CDC Guide for Linux, UNIX, and Windows

Informatica PowerExchange CDC Guide for Linux, UNIX, and Windows
Version 9 .0
December 2009
Copyright (c) 1998-2009 Informatica. All rights reserved.
This software and documentation contain proprietary information of Informatica Corporation and are provided under a license agreement
containing restrictions on use and disclosure and are also protected by copyright law. Reverse engineering of the software is prohibited.
No part of this document may be reproduced or transmitted in any form, by any means (electronic, photocopying, recording or otherwise)
without prior consent of Informatica Corporation. This Software may be protected by U.S. and/or international Patents and other Patents
Pending.
Use, duplication, or disclosure of the Software by the U.S. Government is subject to the restrictions set forth in the applicable software
license agreement and as provided in DFARS 227.7202-1(a) and 227.7702-3(a) (1995), DFARS 252.227-7013 ©(1)(ii) (OCT 1988), FAR
12.212(a) (1995), FAR 52.227-19, or FAR 52.227-14 (ALT III), as applicable.
The information in this product or documentation is subject to change without notice. If you find any problems in this product or
documentation, please report them to us in writing.
Informatica, Informatica Platform, Informatica Data Services, PowerCenter, PowerCenterRT, PowerCenter Connect, PowerCenter Data
Analyzer, PowerExchange, PowerMart, Metadata Manager, Informatica Data Quality, Informatica Data Explorer, Informatica B2B Data
Transformation, Informatica B2B Data Exchange and Informatica On Demand are trademarks or registered trademarks of Informatica
Corporation in the United States and in jurisdictions throughout the world. All other company and product names may be trade names or
trademarks of their respective owners.
Portions of this software and/or documentation are subject to copyright held by third parties, including without limitation: Copyright
DataDirect Technologies. All rights reserved. Copyright © Sun Microsystems. All rights reserved. Copyright © RSA Security Inc. All Rights
Reserved. Copyright © Ordinal Technology Corp. All rights reserved.Copyright © Aandacht c.v. All rights reserved. Copyright Genivia, Inc.
All rights reserved. Copyright 2007 Isomorphic Software. All rights reserved. Copyright © Meta Integration Technology, Inc. All rights
reserved. Copyright © Intalio. All rights reserved. Copyright © Oracle. All rights reserved. Copyright © Adobe Systems Incorporated. All
rights reserved. Copyright © DataArt, Inc. All rights reserved. Copyright © ComponentSource. All rights reserved. Copyright © Microsoft
Corporation. All rights reserved. Copyright © Rouge Wave Software, Inc. All rights reserved. Copyright © Teradata Corporation. All rights
reserved. Copyright © Yahoo! Inc. All rights reserved. Copyright © Glyph & Cog, LLC. All rights reserved.
This product includes software developed by the Apache Software Foundation (http://www.apache.org/), and other software which is
licensed under the Apache License, Version 2.0 (the "License"). You may obtain a copy of the License at http://www.apache.org/licenses/
LICENSE-2.0. Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS
IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific
language governing permissions and limitations under the License.
This product includes software which was developed by Mozilla (http://www.mozilla.org/), software copyright The JBoss Group, LLC, all
rights reserved; software copyright © 1999-2006 by Bruno Lowagie and Paulo Soares and other software which is licensed under the
GNU Lesser General Public License Agreement, which may be found at http://www.gnu.org/licenses/lgpl.html. The materials are provided
free of charge by Informatica, "as-is", without warranty of any kind, either express or implied, including but not limited to the implied
warranties of merchantability and fitness for a particular purpose.
The product includes ACE(TM) and TAO(TM) software copyrighted by Douglas C. Schmidt and his research group at Washington
University, University of California, Irvine, and Vanderbilt University, Copyright ( ©) 1993-2006, all rights reserved.
This product includes software developed by the OpenSSL Project for use in the OpenSSL Toolkit (copyright The OpenSSL Project. All
Rights Reserved) and redistribution of this software is subject to terms available at http://www.openssl.org.
This product includes Curl software which is Copyright 1996-2007, Daniel Stenberg, <daniel@haxx.se>. All Rights Reserved.
Permissions and limitations regarding this software are subject to terms available at http://curl.haxx.se/docs/copyright.html. Permission to
use, copy, modify, and distribute this software for any purpose with or without fee is hereby granted, provided that the above copyright
notice and this permission notice appear in all copies.
The product includes software copyright 2001-2005 ( ©) MetaStuff, Ltd. All Rights Reserved. Permissions and limitations regarding this
software are subject to terms available at http://www.dom4j.org/ license.html.
The product includes software copyright © 2004-2007, The Dojo Foundation. All Rights Reserved. Permissions and limitations regarding
this software are subject to terms available at http:// svn.dojotoolkit.org/dojo/trunk/LICENSE.
This product includes ICU software which is copyright International Business Machines Corporation and others. All rights reserved.
Permissions and limitations regarding this software are subject to terms available at http://source.icu-project.org/repos/icu/icu/trunk/
license.html.
This product includes software copyright © 1996-2006 Per Bothner. All rights reserved. Your right to use such materials is set forth in the
license which may be found at http://www.gnu.org/software/ kawa/Software-License.html.
This product includes OSSP UUID software which is Copyright © 2002 Ralf S. Engelschall, Copyright © 2002 The OSSP Project
Copyright © 2002 Cable & Wireless Deutschland. Permissions and limitations regarding this software are subject to terms available at
http://www.opensource.org/licenses/mit-license.php.
This product includes software developed by Boost (http://www.boost.org/) or under the Boost software license. Permissions and
limitations regarding this software are subject to terms available at http:/ /www.boost.org/LICENSE_1_0.txt.
This product includes software copyright © 1997-2007 University of Cambridge. Permissions and limitations regarding this software are
subject to terms available at http://www.pcre.org/license.txt.
This product includes software copyright © 2007 The Eclipse Foundation. All Rights Reserved. Permissions and limitations regarding this
software are subject to terms available at http:// www.eclipse.org/org/documents/epl-v10.php.
This product includes software licensed under the terms at http://www.tcl.tk/software/tcltk/license.html, http://www.bosrup.com/web/
overlib/?License, http://www.stlport.org/doc/license.html, http://www.asm.ow2.org/license.html, http://www.cryptix.org/LICENSE.TXT,
http://hsqldb.org/web/hsqlLicense.html, http://httpunit.sourceforge.net/doc/license.html, http://jung.sourceforge.net/license.txt , http://
www.gzip.org/zlib/zlib_license.html, http://www.openldap.org/software/release/license.html, http://www.libssh2.org, http://slf4j.org/
license.html, and http://www.sente.ch/software/OpenSourceLicense.htm.
This product includes software licensed under the Academic Free License (http://www.opensource.org/licenses/afl-3.0.php), the Common
Development and Distribution License (http://www.opensource.org/licenses/cddl1.php) the Common Public License (http://
www.opensource.org/licenses/cpl1.0.php) and the BSD License (http://www.opensource.org/licenses/bsd-license.php).
This product includes software copyright © 2003-2006 Joe WaInes, 2006-2007 XStream Committers. All rights reserved. Permissions and
limitations regarding this software are subject to terms available at http://xstream.codehaus.org/license.html. This product includes
software developed by the Indiana University Extreme! Lab. For further information please visit http://www.extreme.indiana.edu/.
This Software is protected by U.S. Patent Numbers 5,794,246; 6,014,670; 6,016,501; 6,029,178; 6,032,158; 6,035,307; 6,044,374;
6,092,086; 6,208,990; 6,339,775; 6,640,226; 6,789,096; 6,820,077; 6,823,373; 6,850,947; 6,895,471; 7,117,215; 7,162,643; 7,254,590; 7,
281,001; 7,421,458; and 7,584,422, international Patents and other Patents Pending..
DISCLAIMER: Informatica Corporation provides this documentation "as is" without warranty of any kind, either express or implied,
including, but not limited to, the implied warranties of non-infringement, merchantability, or use for a particular purpose. Informatica
Corporation does not warrant that this software or documentation is error free. The information provided in this software or documentation
may include technical inaccuracies or typographical errors. The information in this software and documentation is subject to change at
any time without notice.
NOTICES
This Informatica product (the “Software”) includes certain drivers (the “DataDirect Drivers”) from DataDirect Technologies, an operating company of Progress Software
Corporation (“DataDirect”) which are subject to the following terms and conditions:
1. THE DATADIRECT DRIVERS ARE PROVIDED “AS IS” WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
2. IN NO EVENT WILL DATADIRECT OR ITS THIRD PARTY SUPPLIERS BE LIABLE TO THE END-USER CUSTOMER FOR ANY DIRECT, INDIRECT,
INCIDENTAL, SPECIAL, CONSEQUENTIAL OR OTHER DAMAGES ARISING OUT OF THE USE OF THE ODBC DRIVERS, WHETHER OR NOT INFORMED OF
THE POSSIBILITIES OF DAMAGES IN ADVANCE. THESE LIMITATIONS APPLY TO ALL CAUSES OF ACTION, INCLUDING, WITHOUT LIMITATION, BREACH
OF CONTRACT, BREACH OF WARRANTY, NEGLIGENCE, STRICT LIABILITY, MISREPRESENTATION AND OTHER TORTS.
Part Number: PWX-CCl-900-0001

Table of Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi
Informatica Resources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi
Informatica Customer Portal. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi
Informatica Documentation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi
Informatica Web Site. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi
Informatica How-To Library. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Informatica Knowledge Base. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Informatica Multimedia Knowledge Base. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Informatica Global Customer Support. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Part I: PowerExchange CDC Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Chapter 1: Change Data Capture Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

PowerExchange CDC Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Change Data Capture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Change Data Extraction and Apply. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
PowerExchange CDC Data Sources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
DB2 for Linux, UNIX, and Windows Data Sources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Microsoft SQL Server Data Sources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Oracle Data Sources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
i5/OS and z/OS Data Sources with Offload Processing. . . . . . . . . . . . . . . . . . . . . . . . . . 5
PowerExchange CDC Components. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
PowerExchange Listener. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
PowerExchange Logger for Linux, UNIX, and Windows. . . . . . . . . . . . . . . . . . . . . . . . . . 6
PowerExchange Navigator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
PowerExchange Integration with PowerCenter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
PowerExchange CDC Architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Summary of CDC Implementation Tasks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Part II: PowerExchange CDC Components. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Chapter 2: PowerExchange Listener. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

PowerExchange Listener Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Customizing the dbmover.cfg File for CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
CAPI_CONNECTION Statements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Starting the PowerExchange Listener. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Stopping the PowerExchange Listener. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Displaying Active PowerExchange Listener Tasks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Table of Contents i
Chapter 3: PowerExchange Logger for Linux, UNIX, and Windows. . . . . . . . . . . . . 19
PowerExchange Logger Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
PowerExchange Logger Tasks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
PowerExchange Logger Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
CDCT File. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
PowerExchange Logger Log Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Checkpoint Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Cache Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Lock Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Message Log Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
File Switches. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
PowerExchange Logger Operational Modes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Continuous Mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Batch Mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
PowerExchange Logger Considerations on Linux and UNIX. . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
PowerExchange Logger Memory Requirement on Linux or UNIX. . . . . . . . . . . . . . . . . . . 27
Running the PowerExchange Logger in Background Mode. . . . . . . . . . . . . . . . . . . . . . . 27
Configuring the PowerExchange Logger. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Enabling a Capture Registration for PowerExchange Logger Use. . . . . . . . . . . . . . . . . . . 27
Customizing the PowerExchange Logger Configuration File. . . . . . . . . . . . . . . . . . . . . . 28
Customizing dbmover.cfg for the PowerExchange Logger. . . . . . . . . . . . . . . . . . . . . . . . 43
Using PowerExchange Logger Group Definitions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Starting the PowerExchange Logger. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
PWXCCL Syntax and Parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
How the PowerExchange Logger Determines the Start Point for a Cold Start. . . . . . . . . . . 48
Cold Starting the PowerExchange Logger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Managing the PowerExchange Logger. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Commands for Controlling and Stopping PowerExchange Logger Processing. . . . . . . . . . . 49
Assessing PowerExchange Logger Performance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Maintaining the PowerExchange Logger CDCT File and Log Files. . . . . . . . . . . . . . . . . . 53
Backing Up PowerExchange Logger Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Re-creating the CDCT File After a Failure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Part III: PowerExchange CDC Data Sources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Chapter 4: DB2 for Linux, UNIX, and Windows Change Data Capture. . . . . . . . . . . 56
DB2 for Linux, UNIX, and Windows CDC Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Planning for DB2 CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Prerequisites. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Required User Authority. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
CDC Restrictions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Configuring DB2 for CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
ii Table of Contents
Configuring PowerExchange for DB2 CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Configuring PowerExchange CDC without the PowerExchange Logger. . . . . . . . . . . . . . . 59
Configuring PowerExchange CDC with the PowerExchange Logger. . . . . . . . . . . . . . . . . 60
Creating the Capture Catalog Table. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Initializing the Capture Catalog Table. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Customizing dbmover.cfg for DB2 CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Using a DB2 Data Map. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Task Flow for DB2 Data Map Use. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Managing DB2 CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Stopping DB2 CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Changing a DB2 Source Table Definition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Reconfiguring a Partitioned Database or Database Partition Group. . . . . . . . . . . . . . . . . . 67
DB2 for Linux, UNIX, and Windows CDC Troubleshooting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Workaround for SQL1224 Error on AIX. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
IBM APARs for Specific Issues. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Chapter 5: Microsoft SQL Server Change Data Capture. . . . . . . . . . . . . . . . . . . . . . 70

Microsoft SQL Server CDC Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Planning for SQL Server CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
SQL Server CDC Prerequisites. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Required User Authority for SQL Server CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Datatypes Supported for SQL Server CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
SQL Server CDC Restrictions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Configuring SQL Server for CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Configuring PowerExchange for SQL Server CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Configuring PowerExchange CDC without the PowerExchange Logger. . . . . . . . . . . . . . . 74
Configuring PowerExchange CDC with the PowerExchange Logger. . . . . . . . . . . . . . . . . 75
Customizing dbmover.cfg for SQL Server CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Managing SQL Server CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Disabling Publication of Change Data for a SQL Server Source. . . . . . . . . . . . . . . . . . . . 78
Changing a SQL Server Source Table Definition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Chapter 6: Oracle Change Data Capture with Oracle LogMiner. . . . . . . . . . . . . . . . 80

Overview of Oracle LogMiner CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Planning for Oracle LogMiner CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Requirements and Restrictions for Oracle LogMiner CDC. . . . . . . . . . . . . . . . . . . . . . . . 81
Datatypes Supported for Oracle LogMiner CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
SQL*Loader Restrictions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
Performance Considerations for Oracle LogMiner CDC. . . . . . . . . . . . . . . . . . . . . . . . . 83
Oracle Configuration for LogMiner CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Configuration Script Files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Configuring Oracle for LogMiner CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Configuration in an Oracle RAC Environment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Table of Contents iii

PowerExchange Configuration for Oracle LogMiner CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Configuring Oracle LogMiner CDC without the PowerExchange Logger. . . . . . . . . . . . . . . 88
Configuring Oracle LogMiner CDC with the PowerExchange Logger. . . . . . . . . . . . . . . . . 89
Customizing dbmover.cfg for Oracle LogMiner CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . 90
Management of Oracle LogMiner CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
Stopping Oracle LogMiner CDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
Changing a Source Table Definition Used in Oracle LogMiner CDC. . . . . . . . . . . . . . . . . 102
Part IV: Change Data Extraction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Chapter 7: Introduction to Change Data Extraction. . . . . . . . . . . . . . . . . . . . . . . . . 105

Change Data Extraction Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Extraction Modes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
PowerExchange-Generated Columns in Extraction Maps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Restart Tokens and the Restart Token File. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Generating Restart Tokens. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Restart Token File. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Recovery and Restart Processing for CDC Sessions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
PowerCenter Recovery Tables for Relational Targets. . . . . . . . . . . . . . . . . . . . . . . . . . 111
PowerCenter Recovery Files for Nonrelational Targets. . . . . . . . . . . . . . . . . . . . . . . . . 112
Application Names. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Restart Processing for CDC Sessions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Group Source Processing in PowerExchange. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
Using Group Source with Nonrelational Sources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
Using Group Source with CDC Sources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Commit Processing with PWXPC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Controlling Commit Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Maximum and Minimum Rows per Commit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Target Latency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Examples of Commit Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Offload Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
CDC Offload Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Multithreaded Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Chapter 8: Extracting Change Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

Overview of Extracting Change Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Task Flow for Extracting Change Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Testing a Change Data Extraction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Configuring PowerCenter CDC Sessions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Changing Default Values for Session and Connection Attributes. . . . . . . . . . . . . . . . . . . 128
Configuring Application Connection Attributes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Creating Restart Tokens for Extractions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Displaying Restart Tokens. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
iv Table of Contents
Configuring the Restart Token File. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
Restart Token File Statements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
Restart Token File - Example. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Chapter 9: Managing Change Data Extractions. . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

Starting PowerCenter CDC Sessions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Cold Start Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .141
Warm Start Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Recovery Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Stopping PowerCenter CDC Sessions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Stop Command Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Terminating Conditions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Changing PowerCenter CDC Sessions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
Examples of Creating a Restart Point. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .144
Recovering PowerCenter CDC Sessions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .146
Example of Session Recovery. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .146
Chapter 10: Monitoring and Tuning Options. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148

Monitoring Change Data Extractions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .148
Monitoring CDC Sessions in PowerExchange. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .148
Monitoring CDC Sessions in PowerCenter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .151
Tuning Change Data Extractions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .154
Using PowerExchange Parameters to Tune CDC Sessions. . . . . . . . . . . . . . . . . . . . . .155
Using Connection Options to Tune CDC Sessions . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
CDC Offload and Multithreaded Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .159
Planning for CDC Offload and Multithreaded Processing. . . . . . . . . . . . . . . . . . . . . . . .160
Enabling Offload and Multithreaded Processing for CDC Sessions. . . . . . . . . . . . . . . . . 161
Configuring PowerExchange to Capture Change Data on a Remote System. . . . . . . . . . . 162
Extracting Change Data Captured on a Remote System. . . . . . . . . . . . . . . . . . . . . . . .168
Configuration File Examples for CDC Offload Processing. . . . . . . . . . . . . . . . . . . . . . . 168
Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
Table of Contents v
Preface
This guide describes how to configure, implement, and manage PowerExchange Change Data Capture (CDC) on
Linux, UNIX, and Windows systems.
This guide applies to the CDC option of the following PowerExchange products:
¨ PowerExchange for DB2 ® for Linux®, UNIX®, and Windows®
¨ PowerExchange for Oracle ®
¨ PowerExchange for SQL Server ®
Note: If you use the offloading feature, some PowerExchange CDC processing for DB2 for i5/OS data sources and
z/OS data sources can also run on Linux, UNIX, or Windows.
Before implementing change data capture, verify that you have installed the required PowerExchange components.
Informatica Resources
Informatica Customer Portal

As an Informatica customer, you can access the Informatica Customer Portal site at http://my.informatica.com. The
site contains product information, user group information, newsletters, access to the Informatica customer support
case management system (ATLAS), the Informatica How-To Library, the Informatica Knowledge Base, the
Informatica Multimedia Knowledge Base, Informatica Documentation Center, and access to the Informatica user
community.
Informatica Documentation
The Informatica Documentation team takes every effort to create accurate, usable documentation. If you have
questions, comments, or ideas about this documentation, contact the Informatica Documentation team through
email at infa_documentation@informatica.com. We will use your feedback to improve our documentation. Let us
know if we can contact you regarding your comments.
The Documentation team updates documentation as needed. To get the latest documentation for your product,
navigate to the Informatica Documentation Center from http://my.informatica.com.
Informatica Web Site

You can access the Informatica corporate web site at http://www.informatica.com. The site contains information
about Informatica, its background, upcoming events, and sales offices. You will also find product and partner
vi
information. The services area of the site includes important information about technical support, training and
education, and implementation services.
Informatica How-To Library

As an Informatica customer, you can access the Informatica How-To Library at http://my.informatica.com. The How-
To Library is a collection of resources to help you learn more about Informatica products and features. It includes
articles and interactive demonstrations that provide solutions to common problems, compare features and
behaviors, and guide you through performing specific real-world tasks.
Informatica Knowledge Base

As an Informatica customer, you can access the Informatica Knowledge Base at http://my.informatica.com. Use
the Knowledge Base to search for documented solutions to known technical issues about Informatica products.
You can also find answers to frequently asked questions, technical white papers, and technical tips. If you have
questions, comments, or ideas about the Knowledge Base, contact the Informatica Knowledge Base team through
email at KB_Feedback@informatica.com.
Informatica Multimedia Knowledge Base

As an Informatica customer, you can access the Informatica Multimedia Knowledge Base at
http://my.informatica.com. The Multimedia Knowledge Base is a collection of instructional multimedia files that help
you learn about common concepts and guide you through performing specific tasks. If you have questions,
comments, or ideas about the Multimedia Knowledge Base, contact the Informatica Knowledge Base team through
email at KB_Feedback@informatica.com.
Informatica Global Customer Support

You can contact a Customer Support Center by telephone or through the WebSupport Service. WebSupport
requires a user name and password. You can request a user name and password at http://my.informatica.com.
Use the following telephone numbers to contact Informatica Global Customer Support:
North America / South America Europe / Middle East / Africa Asia / Australia
Toll Free Toll Free Toll Free

+1 877 463 2435 00 800 4632 4357 Australia: 1 800 151 830
Singapore: 001 800 4632 4357
Standard Rate Standard Rate Standard Rate

Brazil: +55 11 3523 7761 Belgium: +32 15 281 702 India: +91 80 4112 5738
Mexico: +52 55 1168 9763 France: +33 1 41 38 92 26
United States: +1 650 385 5800 Germany: +49 1805 702 702
Netherlands: +31 306 022 797
Spain and Portugal: +34 93 480 3760
United Kingdom: +44 1628 511 445
Preface vii
viii
Part I: PowerExchange CDC
Introduction
This part contains the following chapters:
¨ Change Data Capture Introduction, 2
1
CHAPTER 1
Change Data Capture Introduction

This chapter includes the following topics:
¨ PowerExchange CDC Overview, 2

¨ PowerExchange CDC Data Sources, 4
¨ PowerExchange CDC Components, 6
¨ PowerExchange Integration with PowerCenter, 7
¨ PowerExchange CDC Architecture, 8
¨ Summary of CDC Implementation Tasks, 10
PowerExchange CDC Overview

PowerExchange Change Data Capture (CDC) works in conjunction with PowerCenter to capture changes to data
in source tables and replicate those changes to target tables and files. This guide describes PowerExchange CDC
for relational database sources on Linux, UNIX, or Windows operating systems.
These sources are:
¨ DB2 for Linux, UNIX, and Windows
¨ Microsoft SQL Server on Windows
¨ Oracle on Linux, UNIX, or Windows
After materializing target tables or files with PowerExchange bulk data movement, you can use PowerExchange
CDC to synchronize the targets with their corresponding source tables. Synchronization is faster when you
replicate only the change data rather than all of the data.
The change data replication process consists of following high-level steps:
1. Change data capture. PowerExchange captures change data for the source tables. PowerExchange can
read change data directly from the RDBMS log files or database. Optionally, you can use the PowerExchange
Logger for Linux, UNIX, and Windows to capture change data to its log files.
2. Change data extraction. PowerExchange, in conjunction with PowerCenter, extracts captured change data
for movement to the target.
3. Change data apply. PowerExchange, in conjunction with PowerCenter, transforms and applies the extracted
change data to target tables or files.
2
Change Data Capture
PowerExchange can capture change data directly from DB2 recovery logs, Microsoft SQL Server distribution
databases, or Oracle redo logs. If you use the offloading feature in combination with the PowerExchange Logger
for Linux, UNIX, and Windows, a PowerExchange Logger process can log change data from data sources on an i5/
OS or z/OS system.
If you do not retain database log files long enough for CDC processing to complete, use the PowerExchange
Logger for Linux, UNIX, and Windows. The PowerExchange Logger writes change data to its log files.
PowerExchange can then extract change data from the PowerExchange Logger log files rather than from the
database log files.
For each source table, you must define a capture registration in the PowerExchange Navigator. The capture
registration provides metadata for the columns that are selected for change capture.
PowerExchange captures changes that result from successful SQL INSERT, DELETE, and UPDATE operations.
Depending on the statement type, PowerExchange captures the following data images:
¨ For INSERTS, PowerExchange captures after images only. An after image reflects a row just after an INSERT
operation. PowerExchange passes these changes as INSERTs to PowerCenter.
¨ For DELETEs, PowerExchange captures before images only. A before image reflects a row just prior to the last
DELETE operation. PowerExchange passes these changes as DELETEs to PowerCenter.
¨ For UPDATEs, PowerExchange captures the following image types:
- Both before and after images if you select an image type of “BA” in the CDC application connection attributes
for PowerCenter. PowerExchange passes an UPDATE to PowerCenter as a DELETE of the before-image
data followed by an INSERT of the after-image data.
- After images if you select an image type of “AI” in the CDC application connection attributes. PowerExchange
passes only the after-image data for an updated row, unless you also request before-image data.
PowerExchange passes an UPDATE to PowerCenter as an UPDATE or INSERT.
Change Data Extraction and Apply

PowerExchange works with PowerCenter to extract change data and write it to one or more target tables or files.
The targets can be on the same system as the source or on a different system.
When you create a capture registration for a source table, the PowerExchange Navigator generates a
corresponding extraction map and application name for the extraction. The extraction map describes the columns
for which to extract change data. You can edit the extraction map to remove columns from extraction processing.
Also, you can create alternative extraction maps, each for a subset of the columns that are registered for capture.
For DB2 for Linux, UNIX, and Windows data sources only, you can create a data map if you have user-defined or
multi-field columns for which you want to manipulate data before loading it to the target.
From PowerCenter, you run a CDC workflow and session that extracts and applies change data. To define a data
source in PowerCenter, you can import the extraction map or import the table definition from the source database
through PowerExchange. For DB2 only, you can import a DB2 data map instead of the extraction map. In most
situations, Informatica recommends that you import the extraction map.
Also, you must define a mapping, session, and workflow in PowerCenter. Optionally, you can include
transformations in the mapping to manipulate the change data. When you define a CDC session, you must specify
a connection type. The connection type determines the extraction mode and access method that PowerExchange
uses to extract data.
To extract change data directly from source DB2 or Oracle log files or SQL Server distribution database, you must
use the real-time extraction mode. To extract change data from PowerExchange Logger log files, you can use
PowerExchange CDC Overview 3

either the batch extraction mode or continuous extraction mode. The following table describes these extraction
modes:
Extraction Mode Description
Real-time extraction mode Reads change data directly from the database log files in near real time, on an ongoing
basis. When the PowerExchange Listener receives an extraction request, it pulls the
change data from the log files and transmits the data to PowerCenter for extraction and
apply processing. This mode provides the lowest latency for change data extraction but
potentially the highest impact on system resources.
Batch extraction mode Reads change data from PowerExchange Logger log files that are in a closed state when
an extraction request is made. After processing the log files, the extraction request ends.
This mode provides the highest latency for change data extraction but minimizes the
impact on system resources.
Continuous extraction mode Reads change data continuously from open and closed PowerExchange Logger log files in
near real time. This mode also minimizes database log accesses and the log retention
period that is required for CDC.
To initiate change data extraction and apply processing, run a CDC workflow and session from PowerCenter.
During extraction processing, PowerExchange extracts changes from the change stream in chronological order
based on the unit of work (UOW) end time. PowerExchange passes only the successfully committed changes to
PowerCenter for processing. PowerExchange does not pass ABORTed or UNDO changes. If you are capturing
changes from DB2 recovery logs or Oracle redo logs, changes that were contiguous in the change stream might
not be contiguous in the reconstructed UOW that PowerExchange passes to PowerCenter.
To properly resume extraction processing, PowerExchange maintains restart tokens for each source table. Restart
tokens are used for all extraction modes. To generate current restart tokens, you can use the PowerExchange
Navigator, the special override statement in the restart token file, or the DTLUAPPL utility.
RELATED TOPICS:
¨ “Introduction to Change Data Extraction” on page 105
PowerExchange CDC Data Sources

PowerExchange can capture change data from DB2 and Oracle data sources on Linux, UNIX, or Windows
systems. PowerExchange can also capture change data from Microsoft SQL Server data sources on Windows.
In the PowerExchange Navigator, you must create a capture registration for each source table. The
PowerExchange Navigator generates a corresponding extraction map and application name. You can import the
extraction map into PowerCenter to define the source for extraction and apply processing.
If you use the PowerExchange Logger for Linux, UNIX, and Windows in combination with the offloading feature,
you can also process change data from data sources on i5/OS or z/OS.
DB2 for Linux, UNIX, and Windows Data Sources

PowerExchange captures change data from DB2 recovery log files for the database that contains your source
tables. For CDC to work, archive logging must be active for the database. Also, you must create a
PowerExchange capture catalog table in the source database. The capture catalog table stores information about
the source tables and columns, including DB2 log positioning information.
4 Chapter 1: Change Data Capture Introduction

If you have a source table with user-defined fields or multi-field columns, you can create a data map to manipulate
these fields with expressions. For example, you might want to create data map to manipulate packed data in a
CHAR column. If you create a data map, you must still create a capture registration and merge the data map with
the extraction map that is generated for the capture registration.
RELATED TOPICS:
¨ “DB2 for Linux, UNIX, and Windows Change Data Capture” on page 56
Microsoft SQL Server Data Sources

PowerExchange CDC uses Microsoft SQL Server transactional replication technology to access data in SQL
Server distribution databases. For CDC to work, you must enable SQL Server Replication on the system from
which change data is captured. Also, verify that each source table in the distribution database has a primary key. If
your database has a high volume of change activity, use a distributed server as the host of the distribution
database. When the extraction process runs, the Microsoft SQL Server Agent must also be running.
RELATED TOPICS:
¨ “Microsoft SQL Server Change Data Capture” on page 70
Oracle Data Sources

PowerExchange uses Oracle LogMiner to read change data from Oracle archive logs. Because PowerExchange
reads data from Oracle archive logs, you must run Oracle in ARCHIVELOG mode. Also, PowerExchange requires
a copy of the Oracle online catalog in the archive logs to determine restart points for change data extraction
processing.
If you have Oracle Version 10g Release 2 or later, PowerExchange supports CDC in Oracle Real Application
Cluster (RAC) environments. In a RAC, the Oracle archive logs for all Oracle instances in the RAC must reside on
shared disk storage for PowerExchange to access them.
RELATED TOPICS:
¨ “Oracle Change Data Capture with Oracle LogMiner” on page 80
i5/OS and z/OS Data Sources with Offload Processing

You can use CDC offload processing in combination with the PowerExchange Logger for Linux, UNIX, and
Windows to log change data from data sources on systems other than the system where the PowerExchange
Logger runs.
With offload processing, a PowerExchange Logger process on Linux, UNIX, and Windows can log change data
from i5/OS and z/OS systems as well as from other Linux, UNIX, or Windows systems. For example, a
PowerExchange Logger process can log change data from a DB2 instance on z/OS.
RELATED TOPICS:
¨ “CDC Offload and Multithreaded Processing” on page 159
PowerExchange CDC Data Sources 5

PowerExchange CDC Components
The following PowerExchange components are used for change data capture (CDC):
¨ PowerExchange Listener. Required, unless PowerExchange and the PowerCenter Integration Service are
installed on the same physical machine.
¨ PowerExchange Logger for Linux, UNIX, and Windows. Optional.
¨ PowerExchange Navigator. Required.
Note: The PowerExchange Condense component has been deprecated in PowerExchange Version 8.6.1.
Although PowerExchange 8.6.1 tolerates continued use of PowerExchange Condense for partial condense
processing, Informatica recommends that you migrate to the PowerExchange Logger. The PowerExchange Logger
replaces PowerExchange Condense. Future PowerExchange versions will require migration to the
PowerExchange Logger.
PowerExchange Listener
The PowerExchange Listener manages capture registrations and extraction maps for all CDC data sources. It also
manages data maps if you create any for DB2 for Linux, UNIX, and Windows tables. The PowerExchange Listener
maintains this information in the following files:
¨ CCT file for capture registrations
¨ CAMAPS directory for extraction maps
¨ DATAMAPS directory for DB2 data maps
The PowerExchange Listener also handles PowerCenter extraction requests for both change data replication and
bulk data movement.
When you create, edit, or delete capture registrations or extraction maps in the PowerExchange Navigator, the
PowerExchange Navigator uses the location value in the registration group and extraction group to contact the
PowerExchange Listener. This location corresponds to a NODE statement in the dbmover.cfg file. For example,
when you open a registration group for a RDBMS instance, the PowerExchange Navigator communicates with the
PowerExchange Listener to get all capture registrations defined for that instance.
A PowerExchange Listener is not required if PowerExchange and the PowerCenter Integration Service run on the
same physical machine.
RELATED TOPICS:
¨ “PowerExchange Listener” on page 12
PowerExchange Logger for Linux, UNIX, and Windows

The PowerExchange Logger for Linux, UNIX, or Windows captures change data from DB2 recovery logs, Oracle
redo logs, or a SQL Server distribution database and writes that data to PowerExchange Logger log files. Use of
the PowerExchange Logger is optional. To use the PowerExchange Logger, run one PowerExchange Logger
process for each database type and instance. The PowerExchange Logger writes all successful UOWs in
chronological order based on end time to its log files. This practice maintains transactional integrity. You can
extract the change data from the PowerExchange Logger log files in either batch or continuous mode.

Benefits of the PowerExchange Logger include:
¨ Source database overhead is reduced because PowerExchange makes fewer accesses to the source log files
or database to read change data. For Oracle, this overhead reduction can be significant. The PowerExchange
Logger can use only one Oracle LogMiner session to read change data for all extractions that process an
Oracle instance.
¨ You do not need to retain the source RDBMS log files longer than normal for CDC.
¨ PowerExchange does not need to reposition its point in the DB2 or Oracle logs from which to resume reading
data. This feature can significantly reduce restart times.
Tip: For Oracle data sources, Informatica recommends that you run the PowerExchange Logger rather than use
real-time extraction mode. Use continuous extraction mode for near-real-time access to change data. This
configuration enables PowerExchange to use one Oracle LogMiner session for all extractions that process an
Oracle instance. Multiple concurrent LogMiner sessions can significantly degrade performance on the machine
where CDC sessions run, including the performance of real-time extractions.
RELATED TOPICS:
¨ “PowerExchange Logger for Linux, UNIX, and Windows” on page 19
PowerExchange Navigator
The PowerExchange Navigator is the graphical user interface from which you define and manage capture
registrations, extraction maps, and data maps.
You must define a capture registration for each source table. The corresponding extraction map is automatically
generated. For DB2 sources, you can also define data maps if you need to perform column-level processing, such
as adding user-defined columns and building expressions to populate them. You can import the extraction maps
into PowerCenter so that they can be used for moving change data to the target.
Note: If the PowerExchange Navigator is not installed on the same machine as a Microsoft SQL Server data
source, you must install the SQL Server client software on the PowerExchange Navigator machine. The client
software is required because PowerExchange uses SQL Server services when creating capture registrations. For
the same situation with DB2 and Oracle data sources, you do not need the RDBMS client software. Instead, from
the PowerExchange Navigator, you can point to the PowerExchange Listener on the machine that contains the
source DB2 database or Oracle instance.
PowerExchange Integration with PowerCenter

PowerCenter provides transformation and data cleansing functions that you can use in CDC sessions. After
capturing change data, use PowerCenter in conjunction with PowerExchange to extract and transform the change
data and then apply it to one or more targets.
To integrate PowerExchange with PowerCenter, use either the PowerExchange Client for PowerCenter (PWXPC)
or the PowerExchange ODBC drivers in PowerCenter. Informatica recommends that you use PWXPC. PWXPC
provides more functionality, better performance, and better recovery and restart capabilities.
Note: This guide assumes that you use PWXPC.
For more information about PWXPC and the PowerExchange ODBC drivers, see PowerExchange Interfaces for
PowerCenter.
PowerExchange Integration with PowerCenter 7

PowerExchange CDC Architecture
The PowerExchange CDC architecture is sufficiently flexible to handle many change data replication scenarios.
You can use PowerExchange in conjunction with PowerCenter to replicate change data from multiple sources of
the same RDBMS type to multiple targets of different types in a single session.
The targets can be tables or files on the same system as the source or on other systems. The PowerCenter
Integration Service can write data to tables in some RDBMSs as well as to flat files and XML files. If you installed
PowerExchange or PowerExchange (PowerCenter Connect) products that provide connectivity to additional
nonrelational or relational targets, you can also load data to those targets, for example, DB2 for z/OS tables,
VSAM data sets, IMS segments, or WebSphere MQ.
You can run multiple instances of PowerExchange CDC components on a single system. For example, you might
want to run a separate PowerExchange Logger for each source RDBMS to create separate sets of log files for
each RDBMS type.
The following figure shows a simple CDC configuration that uses real-time extraction mode to access change data
directly from the change stream without the PowerExchange Logger:
In this real-time configuration, PowerExchange CDC uses the CAPXRT access method to capture change data
from a SQL Server distribution database, DB2 recovery logs, and Oracle redo logs. When an extraction request
runs, PowerCenter connects to the PowerExchange Call Level Interface (SCLI) to contact the PowerExchange
Listener. The change data is passed to the SCLI and then to the PWXPC CDC Real Time reader. In this manner,
the PowerCenter extraction session pulls the change data that PowerExchange captured. After the PWXPC reader
reads the change data, PowerCenter uses the mapping and workflow that you created to transform the data and
load it to the target. With this configuration, you can replicate change data from multiple sources in the same
database or instance to multiple target tables in a single extraction process.
Note: The Oracle UOW Cleanser reconstruct UOWs from redo logs into complete and consecutive UOWs that are
in chronological order by end time. For DB2 and SQL Server, PowerExchange incorporates the UOW Cleanser
function into the consumer API (CAPI) for extracting changes from the data source.

The following figure shows a CDC configuration that uses the PowerExchange Logger in both batch extraction
mode and continuous extraction mode:
In this configuration, the PowerExchange Logger captures change data from the change stream for SQL Server,
Oracle, and DB2 tables and writes that data to its log files. After the data is in the PowerExchange log files, the
source RDBMS log files can be deleted, if necessary. When an extraction session runs, PWXPC contacts the
PowerExchange Listener. The PowerExchange Listener reads the PowerExchange Logger log files and calls the
SCLI on the PowerCenter Integration Service machine to transmit the change data to PowerCenter.
For some source tables, PWXPC extracts change data from the PowerExchange Logger log files in batch
extraction mode with the CAPX access method. In this mode, the extraction session stops after it completes
processing the log files. For other source tables, PWXPC extracts change data in continuous mode with the
CAPXRT access method. In this mode, the extraction session extracts change data on an ongoing basis. In
PowerCenter, you can create one source definition and one mapping that covers both extraction modes. However,
batch and continuous extractions must run as separate sessions. For a batch extraction session, use a PWX CDC
Change application connection. For a continuous extraction session, use a PWX CDC Real Time application
connection. For example, you can run batch extractions to replicate change data to targets that need to be
synchronized periodically, and run continuous extractions to replicate change data to targets that need to be
synchronized in near real time. Batch and continuous extraction sessions can run concurrently.
PowerExchange CDC Architecture 9

Summary of CDC Implementation Tasks
After you install PowerExchange, you can configure change data capture and extraction, materialize targets, and
start extraction processing. The following table identifies the tasks for implementing change data capture and
extraction processing for a data source:
Step Task References
Configure and start PowerExchange CDC components
1 Configure parameters in the dbmover.cfg file for the “Customizing the dbmover.cfg File for CDC” on
PowerExchange Listener. page 12
2 Start the PowerExchange Listener on the machine with the “Starting the PowerExchange Listener” on page
source database. 16
3 Perform RDBMS-specific configuration tasks for CDC. - Chapter 4, “DB2 for Linux, UNIX, and Windows
Change Data Capture” on page 56
- Chapter 5, “Microsoft SQL Server Change Data
Capture” on page 70
- Chapter 6, “Oracle Change Data Capture with
Oracle LogMiner” on page 80
4 (Optional) Configure the PowerExchange Logger. “Configuring the PowerExchange Logger” on page
27
5 (Optional) Start the PowerExchange Logger. “Starting the PowerExchange Logger” on page 47
Define data sources for CDC
6 From the PowerExchange Navigator, define and activate PowerExchange Navigator Guide
capture registrations and extraction maps for the data
sources.
7 For DB2 sources that have user-defined or multi-field PowerExchange Navigator Guide
columns that you want to manipulate, create DB2 data
maps.
Materialize targets and start capturing changes
8 Materialize the target from the source. PowerExchange Bulk Data Movement Guide
9 Establish a start point for the extraction. “Restart Tokens and the Restart Token File” on
page 109
Extract and apply change data
10 From PowerCenter, configure mappings, workflows, - PowerExchange Interfaces for PowerCenter

connections, and sessions. Then run the workflow. - PowerCenter Designer Guide
- PowerCenter Workflow Basics Guide

Part II: PowerExchange CDC
Components
¨ PowerExchange Listener, 12
¨ PowerExchange Logger for Linux, UNIX, and Windows, 19
11
CHAPTER 2
PowerExchange Listener
¨ PowerExchange Listener Overview, 12

¨ Customizing the dbmover.cfg File for CDC, 12
¨ Starting the PowerExchange Listener, 16
¨ Stopping the PowerExchange Listener, 16
¨ Displaying Active PowerExchange Listener Tasks, 17
PowerExchange Listener Overview

In a change data capture (CDC) environment, a PowerExchange Listener can provide some or all of the following
services:
¨ Store and manage capture registrations, extraction maps, and data maps for CDC data sources.
¨ Provide captured change data to PowerCenter when you run a PowerCenter CDC session.
¨ Provide captured change data or source table data to the PowerExchange Navigator when you perform a
database row test of an extraction map or a data map.
¨ Interact with other PowerExchange Listeners on other nodes to facilitate communication among the
PowerExchange Navigator, PowerCenter Integration Service, data sources, and any system to which
PowerExchange processing is offloaded.
Customizing the dbmover.cfg File for CDC

You must configure the parameters in the dbmover.cfg file that pertain to CDC processing. This topic describes
the key CDC parameters that are common to the PowerExchange source RDBMSs on Linux, UNIX, or Windows.
The PowerExchange Listener uses these dbmover.cfg parameters to perform the following functions:
¨ Connect to source RDBMS databases and objects to capture change data.
¨ Determine the directory in which to store capture registrations, extraction maps, and PowerExchange Logger
log files.
¨ Connect to the system with the PowerExchange Logger log files to extract change data.
12
The following table describes the key dbmover.cfg statements that are required for CDC:
Statement Description
CAPI_CONNECTION A named set of parameters that the PowerExchange Consumer API (CAPI) uses to connect to the
change stream and control extraction processing. A CAPI connection is specific to a data source
type. You can define up to eight CAPI_CONNECTION statements in a DBMOVER configuration
file for the same data source type or different data source types. Use the CAPI_SRC_DFLT
parameter to indicate a default CAPI_CONNECTION for a data source type.
PowerExchange requires a connection statement for real-time extraction mode and continuous
extraction mode.
For real-time extraction, PowerExchange uses a source-specific type of CAPI_CONNECTION
statement, such as MSQL, ORCL, and UDB. For more information, see the section for your
source type.
For continuous extraction from PowerExchange Logger log files, PowerExchange CDC uses the
CAPX CAPI_CONNECTION statement.
CAPI_SRC_DFLT The CAPI_CONNECTION statement that PowerExchange uses by default for a specific data
source type when no CAPI connection override is supplied. If you define multiple
CAPI_CONNECTION statements for a data source, you can identify one of them as the default.
Syntax is:
CAPI_SRC_DFLT=(source_type,capi_connection_name)
Where:
- source_type is one of the following source database types: MSS for Microsoft SQL Server,
ORA for Oracle, or UDB for DB2 for Linux, UNIX, and Windows.
- capi_connection_name is the unique name of the CAPI_CONNECTION statement that you
want to use as the default statement.
You can specify a CAPI_SRC_DFLT statement for each source database type.
You can override the default CAPI_CONNECTION with another defined CAPI_CONNECTION in
multiple ways.
CAPT_PATH Path to the local directory that stores the following files for CDC:
- CCT file, which contains capture registrations
- CDEP file, which contains application names for PowerCenter extractions that use ODBC
connections, if any
- CDCT file, which contains information about PowerExchange Logger log files if you use the
PowerExchange Logger
This directory can be a directory that you created specifically for these files or another existing
directory. Informatica recommends that you use a unique directory name to separate these CDC
objects from the PowerExchange code. This practice makes migrating to a new PowerExchange
version easier.
Default is the PowerExchange installation directory.
CAPT_XTRA Path to the local directory that stores extraction maps.

This directory can be a directory that you created specifically for these files or another existing
directory. Informatica recommends that you use a unique directory name to separate these CDC
objects from the PowerExchange code. This practice makes migrating to a new PowerExchange
version easier.
Default is the PowerExchange installation directory.
RELATED TOPICS:
¨ “CAPX CAPI_CONNECTION Parameters” on page 14
Customizing the dbmover.cfg File for CDC 13

CAPI_CONNECTION Statements
PowerExchange requires that you define CAPI_CONNECTION statements in the dbmover.cfg file on any Linux,
UNIX, or Windows system where PowerExchange captures or extracts change data. PowerExchange uses the
parameters that you specify in the CAPI_CONNECTION statements to connect to the change stream and to
customize capture and extraction processing.
For each data source, you must define one of the following source-specific types of CAPI_CONNECTION
statements:
¨ For Microsoft SQL Server, an MSQL CAPI_CONNECTION

¨ For Oracle, an ORCL CAPI_CONNECTION and a UOW CAPI_CONNECTION for the UOW Cleanser
¨ For DB2 for Linux, UNIX, and Windows, a UDB CAPI_CONNECTION
If you use continuous extraction mode to extract change data from PowerExchange Logger log files, you must also
define a CAPX CAPI_CONNECTION statement.
You can specify up to eight CAPI_CONNECTION statements in a dbmover.cfg file. You can identify one of the
statements as the overall default. If you define multiple CAPI_CONNECTION statements for the same source type,
you can identify one of these statements as the source-specific default. In addition to or in lieu of defaults, you can
define specific CAPI_CONNECTION overrides in multiple ways. The order of precedence that PowerExchange
uses to determine which CAPI_CONNECTION statement to use is described in the PowerExchange Reference
Manual.
Note: When you extract change data, PowerExchange uses CAPI_CONNECTION statements to connect to the
change stream for the data source. To perform database row tests for data sources that are defined by capture
registrations local to the PowerExchange Navigator, you must specify the appropriate CAPI_CONNECTION
statements on the PowerExchange Navigator machine. Otherwise, you do not need to specify
CAPI_CONNECTION statements to perform database row tests.
RELATED TOPICS:
¨ “DB2 for Linux, UNIX, and Windows CAPI_CONNECTION Parameters” on page 62
¨ “Microsoft SQL Server CAPI_CONNECTION Parameters” on page 76
¨ “ORCL CAPI_CONNECTION Statement” on page 92
¨ “UOWC CAPI_CONNECTION Statement” on page 99
CAPX CAPI_CONNECTION Parameters

The CAPX CAPI_CONNECTION statement specifies the Consumer API (CAPI) parameters needed for continuous
extraction of change data from PowerExchange Logger for Linux, UNIX, and Windows log files.
Operating Linux, UNIX, and

Systems: Windows
Required: Yes for continuous
extraction mode
Syntax:
CAPI_CONNECTION=(
[DLLTRACE=trace_id,]
NAME=name,
[TRACE=trace,]
TYPE=(CAPX,
DFLTINST=collection_id,
[FILEWAIT=seconds,]
[RSTRADV=seconds]
14 Chapter 2: PowerExchange Listener

)
)
Parameters:
Enter the following parameters:
DLLTRACE=trace_id
Optional. User-defined name of the TRACE statement that activates internal DLL tracing for this CAPI.
Specify this parameter only at the direction of Informatica Global Customer Support.
NAME=name
Required. Unique user-defined name for this CAPI_CONNECTION statement.
Maximum length is eight alphanumeric characters.
TRACE=trace
Optional. User-defined name of the TRACE statement that activates the common CAPI tracing. Specify this
parameter only at the direction of Informatica Global Customer Support.
TYPE=(CAPX, ... )
Required. Type of CAPI_CONNECTION statement. For continuous extraction mode, this value must be CAPX.
DFLTINST=collection_id
Required. A source identifier, sometimes called the instance name or collection identifier, that is defined
in capture registrations. This value must match the instance or database name that is displayed in the
Resource Inspector of the PowerExchange Navigator for the registration group that contains the capture
registrations.
FILEWAIT=seconds
Optional. Time interval, in seconds, that PowerExchange waits before checking for new PowerExchange
Logger log files.
Valid values are from 1 through 86400.
Default is 1.
RSTRADV=nnnnn
Time interval, in seconds, that PowerExchange waits before advancing restart and sequence tokens for a
registered data source during periods when UOWs do not include any changes of interest for the data
source. When the wait interval expires, PowerExchange returns the next committed "empty UOW," which
includes only updated restart information.
The wait interval is reset to 0 when PowerExchange completes processing a UOW that includes changes
of interest or returns an empty UOW because the wait interval expired without any changes of interest
having been received.
For example, if you specify 5, PowerExchange waits 5 seconds after it completes processing the last
UOW or after the previous wait interval expires. Then PowerExchange returns the next committed empty
UOW that includes the updated restart information and resets the wait interval to 0.
If RSTRADV is not specified, PowerExchange does not advance restart and sequence tokens for a
registered source during periods when no changes of interest are received. In this case, when
PowerExchange warm starts, it reads all changes, including those not of interest for CDC, from the
restart point.
Valid values are 0 through 86400. No default is provided.
Customizing the dbmover.cfg File for CDC 15

Warning: A value of 0 can degrade performance because PowerExchange returns an empty UOW after
each UOW processed.
Starting the PowerExchange Listener

To start the PowerExchange Listener, you can run the dtllst program or use other system-specific methods.
On a Linux or UNIX system, use one of the following methods:
¨ Enter dtllst at the command line to run the PowerExchange Listener in foreground mode:
dtllst node1 [config=directory/myconfig_file] [license=directory/mylicense_key_file]
Include the optional config and license parameters if you want to specify configuration and license key files that
override the original dbmover.cfg and license.key files.
You can add an ampersand (&) at the end to run the PowerExchange Listener in background mode and add
the prefix "nohup" at the beginning to run the PowerExchange Listener persistently:
nohup dtllst node1 [config=directory/myconfig_file] [license=directory/mylicense_key_file] &
¨ Run the startlst script, which was installed with PowerExchange. This script deletes the detail.log file and then
starts the PowerExchange Listener.
On a Windows system, use one of the following methods:
¨ Run the PowerExchange Listener as a Windows service, which is the usual practice. To start a
PowerExchange Listener service from the Windows Start menu, click Start > Programs > Informatica
PowerExchange > Start PowerExchange Listener. Alternatively, use the dtllstsi program to enter the start
command from a Windows command prompt:
dtllstsi start “service_name”
¨ Enter dtllst. The syntax is the same as for Linux and UNIX except that the & and nohup operands are not
supported. Your product license must allow this manual mode of PowerExchange Listener operation.
Note: You cannot start the PowerExchange Listener by using the pwxcmd program.
Stopping the PowerExchange Listener

To stop the PowerExchange Listener, use the CLOSE or CLOSE FORCE command. To stop active
PowerExchange Listener tasks, use the STOPTASK command.

The following table describes these commands and the syntax for issuing each command from the command line
against a PowerExchange Listener task that is running in foreground mode:
Command Description Command Line Syntax
CLOSE Stops the PowerExchange Listener after all of the On Linux, UNIX, or Windows:
following subtasks complete: C
- CDC subtasks, which stop at the next commit of
a unit of work (UOW)
- Bulk data movement subtasks
- PowerExchange Listener subtasks
CLOSE FORCE Forces the cancellation of all user subtasks and On Linux or UNIX:
stops the PowerExchange Listener. C F
PowerExchange waits 30 seconds for current user On Windows:
CF
subtasks on the PowerExchange Listener to
complete. Then PowerExchange cancels any
remaining user subtasks and stops the
PowerExchange Listener. This command is useful
if you have long-running subtasks on the
PowerExchange Listener.
STOPTASK Stops a PowerExchange Listener task for a On Linux or UNIX:

specific extraction application process. STOPTASK app_name
PowerExchange waits to stop the PowerExchange On Windows:
STOPTASK APPLID=app_name
Listener until either the end UOW or commit
The app_name is the name of an active change
threshold is reached.
data extraction process. You can get this name
from the PWX-00712 messages in the
PowerExchange Listener D (DISPLAY ACTIVE)
command output.
Alternatively, you can use any of the following methods:
¨ On a Linux, UNIX, or Windows system, use the pwxcmd program to issue the close, closeforce, or stoptask
command to a PowerExchange Listener running in foreground or background mode, on the local system or a
remote system. You can issue these pwxcmd commands from the command line or include them in scripts or
batch files.
¨ On a Linux or UNIX system, if the PowerExchange Listener is running in background mode, use the standard
operating system commands to find the PowerExchange Listener process ID and then “kill” that process. A “kill”
operation is similar to a CLOSE operation.
¨ On a Windows system, if the PowerExchange Listener does not respond to a CLOSE FORCE command, press
Ctrl + C once to issue CLOSE or press Ctrl + C twice to issue CLOSE FORCE.
Displaying Active PowerExchange Listener Tasks

You can use the DISPLAY ACTIVE command to display information about each active PowerExchange Listener
task that is running in foreground mode on a Linux, UNIX, or Windows system. This information includes the TCP/
IP address, port number, application name, access type, and status.
On a Linux, UNIX, or Windows system, enter the following command at the command line on the screen where the
PowerExchange Listener task is running in foreground mode:
D
Displaying Active PowerExchange Listener Tasks 17

Alternatively, on a Linux, UNIX, or Windows system, you can issue the pwxcmd listtask command from a
command line, script, or batch file to a PowerExchange Listener running on the local system or a remote system.
The pwxcmd listtask command produces the same output as the DISPLAY ACTIVE command.

CHAPTER 3
PowerExchange Logger for Linux,

UNIX, and Windows
¨ PowerExchange Logger Overview, 19
¨ PowerExchange Logger Tasks, 20
¨ PowerExchange Logger Files, 21
¨ File Switches, 25
¨ PowerExchange Logger Operational Modes, 25
¨ PowerExchange Logger Considerations on Linux and UNIX, 27
¨ Configuring the PowerExchange Logger, 27
¨ Starting the PowerExchange Logger, 47
¨ Managing the PowerExchange Logger, 49
PowerExchange Logger Overview

The PowerExchange Logger for Linux, UNIX, and Windows captures change data from PowerExchange data
sources and write that data to PowerExchange Logger log files. The PowerExchange Logger writes only the
successful units of work (UOWs) to its log files, in chronological order based on end time.
When a PowerCenter CDC session runs, it extracts change data from the log files instead of from the change
stream.
Note: The PowerExchange Logger for Linux, UNIX, and Windows is similar in function to PowerExchange
Condense on i5/OS or z/OS systems.
The PowerExchange Logger can capture change data from DB2 recovery logs or Oracle redo logs on Linux, UNIX,
or Windows, or from a Microsoft SQL Server distribution database on a Windows. If you use the offloading feature,
a PowerExchange Logger process on Linux, UNIX, or Windows can also process data from data sources on i5/OS
or z/OS systems.
Use the PowerExchange Logger to reduce database overhead due to CDC processing. With the PowerExchange
Logger, PowerExchange accesses the source database fewer times to read change data, which reduces database
I/O. Also, because change data is extracted from the PowerExchange Logger log files, you often do not need to
extend the retention period for source database log files to accommodate CDC processing.
You must run one PowerExchange Logger process for each source type and instance, as defined in a registration
group. The PowerExchange Logger runs in continuous mode or batch mode.
19
When you create capture registrations for data sources, including i5/OS and z/OS data sources for which
processing is offloaded, set the Condense option to Part. The PowerExchange Logger supports only partial
condense processing. For i5/OS or z/OS data sources, if you set the Condense option to Full in capture
registrations, the PowerExchange Logger ignores the registrations and does not process change data from those
sources.
For each PowerExchange Logger process, you must define a configuration file. PowerExchange provides a
sample configuration file named pwxccl.cfg. The configuration file contains parameters for controlling the
PowerExchange Logger and for identifying the source instance. Use the COLL_END_LOG parameter to control
whether the PowerExchange Logger runs in continuous mode or batch mode.
When PowerCenter workflow sessions run, you can extract change data from PowerExchange Logger log files in
batch extraction mode or continuous extraction mode. Do not use real-time extraction mode with the
PowerExchange Logger.
Tip: For Oracle near-real-time CDC, Informatica recommends that you use the PowerExchange Logger and
continuous extraction mode. PowerExchange then uses one Oracle LogMiner session for all extractions that
process an Oracle instance. If you use real-time extraction mode, without the PowerExchange Logger,
PowerExchange starts a separate LogMiner session for each extraction. The use of multiple, concurrent LogMiner
sessions can significantly degrade the performance on the system where LogMiner runs.
RELATED TOPICS:
¨ “PowerExchange Logger Operational Modes” on page 25
¨ “Customizing the PowerExchange Logger Configuration File” on page 28
PowerExchange Logger Tasks

The PowerExchange Logger uses a Controller task with Command Handler and Writer subtasks.
These tasks perform the following functions:
Controller task
Loads parameter settings from the PowerExchange Logger pwxccl.cfg configuration file. Reads the cache file
from the last run to determine if capture registrations have been added or removed, and loads the capture
registrations from the CCT file. After loading this information, the Controller starts the Command Handler
subtask and then the Writer subtask.
Command Handler subtask
Processes PowerExchange Logger commands from various sources, including user stdin and the pwxcmd
program. If the PROMPT parameter is set to Y in the pwxccl.cfg file, the Command Handler waits for the
Writer subtask to initialize before accepting a user command.
Writer subtask
Performs most of the PowerExchange Logger work that uses CPU time. The Writer initializes the CAPI for the
source database, determines the start or restart point in the change stream, reads change data from the
change stream, and writes change data to PowerExchange Logger log files. The Writer also performs
checkpoint processing, writes records to the CDCT file during a file switch, deletes expired CDCT records,
and rolls back CDCT records when you warm start the PowerExchange Logger from an earlier point in time. If
the PROMPT parameter is set to Y in the pwxccl.cfg file, the Writer waits for you to respond to confirmation
prompts before proceeding with a cold start or a rollback of CDCT records.
20 Chapter 3: PowerExchange Logger for Linux, UNIX, and Windows

PowerExchange Logger Files
A PowerExchange Logger process writes information to the CDCT file, checkpoint files, PowerExchange Logger
log files, and PowerExchange message logs.
It also uses cache files and lock files during processing.
CDCT File
The PowerExchange Logger stores information about its log files in the CDCT file. When a PowerCenter CDC
session runs in continuous extraction mode or batch extraction mode, the PowerExchange Listener reads the
CDCT file to determine the PowerExchange Logger log files from which to extract change data.
The PowerExchange Logger creates the CDCT file in the directory that is specified by the CAPT_PATH statement
in the dbmover.cfg file on the system where the PowerExchange Logger runs. If the CAPT_PATH statement is not
specified, the CDCT file is in the directory from which the PowerExchange Logger is invoked.
After a file switch, or the first time the PowerExchange Logger receives change data based on an active capture
registration, the PowerExchange Logger Writer subtask writes keyed records to the CDCT file. These records
contain information about each closed PowerExchange Logger log file, including the log file name, number of
records read, UOW start and end times, whether before images are included, and other control information.
For example, if a log file contains change records for two registration tags, or tables, and you are not using a
group definition file, the following processing occurs:
1. When source data for each table is first received and written to the log file, the Writer subtask writes a
temporary record for the log file, which does not include a registration tag name, to the CDCT file. This
temporary record enables the PowerExchange Logger to retrieve source data for extractions that run in
continuous extraction mode.
2. When a file switch occurs, the Writer subtask writes two keyed records to the CDCT file, one for each of the
registration tags. Each record includes the registration tag name, log file name, and change record count.
3. The PowerExchange Logger then deletes the temporary CDCT records that do not include the registration
tags.
If you use a group definitions file, processing is similar to that in the previous example except that the Writer
subtask writes one temporary record without a registration tag for each log file that received source data. You can
have as many temporary records as groups in the group definition file.
Tip: You can use the PWXUCDCT utility to print information about CDCT records, back up and restore the CDCT
file, re-create the CDCT file based on PowerExchange Logger log files if necessary, and delete expired CDCT
records.
RELATED TOPICS:
¨ “Maintaining the PowerExchange Logger CDCT File and Log Files” on page 53
PowerExchange Logger Log Files

The PowerExchange Logger creates log files for storing change data records when it first encounters changes for
source tables and columns of interest. These source tables and columns must be defined in active capture
registrations.
PowerExchange Logger Files 21

The PowerExchange Logger creates log files based on the EXT_CAPT_MASK parameter in the pwxccl.cfg file.
This parameter specifies a path to the directory where log files are stored and a prefix for the log file names. Log
file names have the following format:
path/prefix.CND.CPyymmdd.Thhmmssnnn
Where:
¨ path/prefix is the EXT_CAPT_MASK value.
¨ yymmdd is the date when the file is created.
¨ hhmmss is a 24-hour time when the file is created.

¨ nnn is a generated sequence number, starting at 001, that makes each file name unique.
The log files remain open until a file switch occurs or the PowerExchange Logger shuts down.
When you run a PowerCenter CDC session in continuous extraction mode or batch extraction mode,
PowerExchange extracts change data from the PowerExchange Logger log files.
RELATED TOPICS:
Checkpoint Files
The PowerExchange Logger creates checkpoint files to store restart tokens and sequence tokens for correctly
resuming CDC processing after a PowerExchange Logger warm start.
The PowerExchange Logger writes information to the checkpoint files each time a file switch occurs or a
SHUTDOWN or SHUTCOND command is issued.
Note: Checkpoint files are not used for PowerExchange Logger cold starts.
The PowerExchange Logger creates checkpoint files based on the CHKPT_BASENAME and CHKPT_NUM
parameters in the pwxccl.cfg file, as follows:
¨ The CHKPT_BASENAME parameter specifies the path to the directory where checkpoint files are stored and a
base file name. Checkpoint file names have the following format:
path/base_name.Vn.ckp
Where:
- path/base_name is the CHKPT_BASENAME value.
- n is a number that the PowerExchange Logger appends to the file name. This number can be a value from 0
to (CHKPT_NUM value - 1).
¨ The CHKPT_NUM parameter specifies the number of checkpoint files. At least two checkpoint files are required.

Checkpoint files are sequential files that have a binary variable-length format. Checkpoint files contain the
following types of records:
Checkpoint Record Description
1 Main record that contains the checkpoint timestamp and restart and sequence tokens. This
information is used to determine the restart point in the change stream for a PowerExchange
Logger warm start.
2 Optional. Uncommitted registrations at the end of a PowerExchange Logger log file that did not
end on a Commit record.
3 Optional. Names of the PowerExchange Logger log files that were closed.
If you need to relocate a PowerExchange Logger configuration, you can copy the checkpoint files to another
machine that has the same integer endian format. However, you cannot copy checkpoint files to a machine that
uses a different integer endian format because the integer fields in checkpoint files that define record length are
platform-dependent.
To display information about checkpoint files, you can use the following methods:
¨ Issue the DISPLAY CHECKPOINTS command to display a message that reports the sequence number and
timestamp of the last checkpoint file written.
¨ Use the PWXUCDCT utility REPORT_CHECKPOINTS command to print a report that provides information
about each checkpoint file, including its timestamp, restart and sequence tokens, reason for the checkpoint,
number of expired CDCT records that were deleted, and number of log files to which data was written.
RELATED TOPICS:
Cache Files
The PowerExchange Logger creates two identical cache files, one of which is a backup, in the
CHKPT_BASENAME directory. The cache files store registration tag names for warm start processing.
When the PowerExchange Logger warm starts, it reads the cache file from its last run to determine if any capture
registrations have been added or removed. If so, the PowerExchange Logger issues message PWX-06119.
Lock Files
During initialization, a PowerExchange Logger process creates lock files to prevent other PowerExchange Logger
processes from accessing the same CDCT file, checkpoint files, and log files concurrently.
As long as the PowerExchange Logger process holds a lock on the lock files, locking is in effect for the resources
for which the lock files were created.
PowerExchange Logger locking works on local disks on Linux, UNIX, or Windows systems. It also works on the
following shared file systems on Linux or UNIX systems:
¨ Veritas Storage Foundation ™ Cluster File System by Symantec
¨ IBM General Parallel File System
¨ EMC Celerra network-attached storage (NAS) with Network File System (NFS) protocol version 3
¨ NetApp NAS with NFS version 3
PowerExchange Logger Files 23

The PowerExchange Logger creates lock files in the following order:
1. A lock file for the CDCT file for a source instance. The PowerExchange Logger generates the lock file name
and location based on the directory that is specified in the CAPT_PATH parameter of the dbmover.cfg file.
2. A lock file for checkpoint files. The PowerExchange Logger generates the lock file name and location based
on the directory and base file name that are specified in the CHKPT_BASENAME parameter of the pwxccl.cfg
file.
3. One of the following lock files:
¨ If you do not use a group definition file, a lock file for PowerExchange Logger log files. The
PowerExchange Logger generates the lock file name and location based the directory and file-name prefix
that are specified in the EXT_CAPT_MASK parameter of the pwxccl.cfg file.
¨ If you use a group definition file, a lock file for each set of the PowerExchange Logger log files that is
defined by the GROUP statements in the group definition file. The PowerExchange Logger generates the
lock file names and locations based on the external_capture_mask parameter in each GROUP statement.
In this case, the PowerExchange Logger ignores the EXT_CAPT_MASK parameter in the pwxccl.cfg file
when creating lock files and processing log files.
Lock file names end with _lockfile.lck. For example, a lock file for the CDCT file could have the name
CDCT_oracoll1_lockfile.lck.
When the PowerExchange Logger process ends, it unlocks the lock files to enable other PowerExchange Logger
processes to access the previously locked resources.
To identify a PowerExchange Logger process that holds a lock, look up the process ID (PID) in the Task Manager
on a Windows system or issue the ps command on a UNIX or Linux system.
Also, the PowerExchange Logger writes messages to the PowerExchange message log that indicate the locking
status. Look for the following key messages:
¨ To verify that lock files are created, look for PWX-25802 messages, such as:
PWX-25802 Process pwxccl.exe pid 5428 locked file C:\capture\captpath\CDCT_instance_lockfile.lck
¨ To verify that lock files are unlocked, look for PWX-25803 messages, such as:
PWX-25803 Process pwxccl.exe pid 5428 unlocked file C:\capture\extcapt\loggerfiles_lockfile.lck
¨ If the PowerExchange Logger process cannot find the lock file that it needs to access some resources, it writes
message PWX-25800:
PWX-25800 Could not find lock file file_name
¨ If a lock file is locked by another process, the PowerExchange Logger process writes some or all of the
following messages, depending on if it can acquire a lock before the maximum retry interval that is specified in
PWX-25814 elapses:
PWX-25804 Error trying to lock PowerExchange Logger files
PWX-25811 File file_name is locked by process process_name pid process_id on host host_name date
date time time
PWX-25812 File file_name is locked by pid process_id start offset length bytes
PWX-25813 No information is available on process which locked file file_name
PWX-25814 Trying to lock file file_name until number seconds elapses
PWX-25815 File file_name is locked by another process and no more waiting is allowed.
If a PowerExchange Logger process ends abnormally with message PWX-25815 and return code 25815, try to
determine the status of the other PowerExchange Logger process that is holding the lock. This other process is
identified in message PWX-25811. For example, the other process might not have completely shut down, or both
processes might be trying to use the same files because of an error in their pwxccl.cfg configuration files.
Message Log Files

The PowerExchange Logger writes messages to the PowerExchange message log file.

By default, on Linux, UNIX, and Windows, this file is named detail.log and is located in the working directory where
the PowerExchange Logger process runs. However, you can optionally specify another directory for
PowerExchange message log files. You can also enable the use of alternative log files.
To specify a unique directory for PowerExchange message log files, include the LOGPATH parameter in the
dbmover.cfg file. Use of this parameter can help you find the PowerExchange message log files more easily.
Also, you can implement alternative logging by specifying the TRACING statement in the dbmover.cfg file. When
alternative logging is enabled, PowerExchange creates a set of alternative log files for each PowerExchange
process, including each PowerExchange Logger process, in a separate directory. When an alternative log file
becomes full, PowerExchange switches to another alternative log file. This automatic rotation of message log files
prevents out-of-space conditions. Also, PowerExchange buffers messages before writing them to the alternative
log files on disk at a specific flush interval. This mode of writing messages can reduce I/O activity on the
alternative log files.
File Switches
When running in continuous mode, the PowerExchange Logger periodically closes its open log files if they contain
data and then opens a new set of log files. This process is called a file switch.
The PowerExchange Logger automatically performs a file switch when the criteria in the following parameters of
the pwxccl.cfg file are met:
¨ FILE_SWITCH_CRIT
¨ FILE_SWITCH_MIN
¨ FILE_SWITCH_VAL
If the open log files do not contain data when the file-switch criteria in these parameters are met, the file switch
does not occur. The PowerExchange Logger waits until the file-switch criteria are met again. If the files still do not
contain data, the PowerExchange Logger continues to check the log files at set intervals. Only when the log files
contain data does the file switch occur.
Also, you can force a file switch by entering the fileswitch command from the command line. Alternatively, on
Linux, UNIX, or Windows, you can send a pwxcmd fileswitch command to a PowerExchange Logger process
running on the local system or a remote system.
RELATED TOPICS:
¨ “Configuring the PowerExchange Logger” on page 27
PowerExchange Logger Operational Modes

A PowerExchange Logger process can operate in continuous mode or batch mode.
To set the operational mode, use the COLL_END_LOG parameter in the pwxccl.cfg file.
RELATED TOPICS:
¨ “File Switches” on page 25
¨ “Extraction Modes” on page 106
File Switches 25
Continuous Mode
In continuous mode, the PowerExchange Logger process runs continuously until you manually stop it.
Consider using continuous mode in the following situations:
¨ You have a database with a high level of change activity that occurs continuously.
¨ You have a database with intermittent activity that occurs at unpredictable intervals.
¨ You want to avoid the overhead of scheduling PowerExchange Logger runs.
¨ You cannot restart the PowerExchange Logger process often enough to keep up with the change volume.
To enable continuous mode, set the COLL_END_LOG parameter to 0.
In continuous mode, each time the Writer subtask completes a logging cycle, the PowerExchange Logger process
is temporarily suspended. The next cycle is triggered by any of the following events:
¨ The wait interval that is defined in the NO_DATA_WAIT parameter of the pwxccl.cfg file elapses.
¨ The CONDENSE command is manually entered at the command line or with the pwxcmd program.
¨ The FILESWITCH command is manually entered at the command line or with the pwxcmd program.
The PowerExchange Logger process continues to run until you enter the SHUTDOWN or SHUTCOND command.
To prevent log files from becoming too large, the PowerExchange Logger process periodically performs a file
switch. Files that are too large can extend restart times for CDC sessions that run in continuous extraction mode
or batch extraction mode.
You can use the NO_DATA_WAIT2 parameter in the pwxccl.cfg file to prevent the PowerExchange Logger from
consuming too much CPU time when PowerExchange is not receiving changes. For example, if you set the
NO_DATA_WAIT2 parameter to 30 seconds, the PowerExchange Logger sleeps for 30 seconds, provided that no
updates are received, and then performs another processing cycle. However, a large NO_DATA_WAIT2 value can
delay processing of a SHUTDOWN command. If you need to reduce the amount of time that the PowerExchange
Logger sleeps on a quiet system, you can adjust the FILE_FLUSH_VAL, FILE_SWITCH_VAL, and
FILE_SWITCH_MIN parameters.
Run the PowerExchange Logger in continuous mode unless you have a specific reason to use batch mode.
Tip: On a Linux or UNIX system, you can run a continuous PowerExchange Logger process in background mode.
Then use the pwxcmd program to send commands to the PowerExchange Logger process that is running in
background mode.
When you run the PowerExchange Logger in continuous mode, you can use either continuous or batch extraction
mode for workflows that extract change data from the PowerExchange Logger log files.
Batch Mode
The PowerExchange Logger process shuts down after the number of seconds in the NO_DATA_WAIT2 parameter
of the pwxccl.cfg file elapse and no data has been received.
Use batch mode in the following situations:
¨ You want to run the PowerExchange Logger on a scheduled basis after batch applications that update the
database complete.
¨ You want to run the PowerExchange Logger manually or for testing.
To enable batch mode, set the COLL_END_LOG parameter to 1 in the pwxccl.cfg file. Also, set NO_DATA_WAIT2
parameter to the number of seconds that PowerExchange waits at the end-of-log for more change data before
shutting down the PowerExchange Logger.
When you run the PowerExchange Logger in batch mode, use batch extraction mode for any workflows that
extract change data from the PowerExchange Logger log files.

PowerExchange Logger Considerations on Linux and
UNIX
If you run the PowerExchange Logger on a Linux or UNIX system, review the requirements for the amount of
memory needed and for running the PowerExchange Logger in background mode.
PowerExchange Logger Memory Requirement on Linux or UNIX

The PowerExchange Logger requires sufficient amounts of main memory and virtual memory to process change
data.
If the memory is not sufficient, PowerExchange writes the error messages PWX-00271 and PWX-00904 to the
PowerExchange message log file when you attempt to start the PowerExchange Logger on Linux or UNIX.
To prevent this problem, use the Linux or UNIX ulimit command to set the size limits for maximum memory and
virtual memory to unlimited. The specific ulimit syntax varies by platform and shell. For more information about this
command, see the documentation for your Linux or UNIX operating system.
Running the PowerExchange Logger in Background Mode

You can run a PowerExchange Logger process in background mode on Linux or UNIX systems.
For background PowerExchange Logger processes, Informatica recommends that you set the COLL_END_LOG
parameter to 0 in the pwxccl.cfg file to run the PowerExchange Logger continuously. Also, accept the default value
of N for the PROMPT parameter. If you specify PROMPT=Y, the PowerExchange Logger ignores this setting and
issues an error message.
To send commands to a PowerExchange Logger process that is running in the background, use the pwxcmd
program. To enable pwxcmd use, define the CONDENSENAME parameter in the pwxccl.cfg file and define the
SVCNODE statement in the dbmover.cfg file.
Configuring the PowerExchange Logger

To configure the PowerExchange Logger, you must define a PowerExchange Logger configuration file for each
source type and instance, as defined in a registration group. Also, verify that the Condense option is set to Part in
the capture registrations for all sources that PowerExchange Logger processes.
If you want the PowerExchange Logger to create separate log files for one or more groups of tables, create a
PowerExchange group definition file that defines groups of capture registrations for the tables.
Enabling a Capture Registration for PowerExchange Logger Use

For the PowerExchange Logger to use a capture registration, the registration must have a status of active and a
Condense setting of Part.
If the PowerExchange Logger does not find any active capture registration, the PowerExchange Logger issues
error message PWX-06427 and ends.
PowerExchange Logger Considerations on Linux and UNIX 27

To enable a capture registration for PowerExchange Logger use:
1. In the PowerExchange Navigator, open the capture registration.

2. In the Resource Inspector, select Active in the Status list.
3. In the Condense list, select Part.
Customizing the PowerExchange Logger Configuration File

Before you start the PowerExchange Logger, configure its parameters in the PowerExchange Logger configuration
file.
PowerExchange provides an example configuration file, named pwxccl.cfg, in the PowerExchange installation
directory that is specified in the PWX_HOME environment variable on Linux or UNIX or PATH environment
variable on Windows. Use this example file as a starting point for your customized file. You can rename the
example file and copy it to another directory. If you do so, you must specify the CS parameter when you start the
PowerExchange Logger to identify the alternative path or file name or both.
If you used the similar PowerExchange Condense feature in an earlier PowerExchange release, you can copy its
dtlca.cfg configuration file and then customize the copy. You might want to add PowerExchange Logger
parameters that PowerExchange Condense did not support. Rename the file to pwxccl.cfg or use the CS
execution parameter. The PowerExchange Logger replaces PowerExchange Condense on Linux, UNIX, and
Windows.
If you specify a parameter value that contains one or more spaces, such as a Windows path, you must enclose the
value in double quotation marks. Make sure that you use straight quotation marks (").
Parameter Descriptions
This topic describes the PowerExchange Logger parameters that you can specify in pwxccl.cfg.

The parameters are:
Parameter Description Valid Values
CAPT_IMAGE Data image type that the PowerExchange Logger - AI for after images.
captures to its log files. The PowerExchange - BA for before and after
Logger can capture after images only or both images.
before and after images of the data. Default is AI.
This image type must be consistent with the
image type delivered to the target during
extraction processing.
If you enter AI for this parameter, the following
limitations apply:
- You cannot extract before images to the target.
- You cannot use DTL_BI columns in extraction
maps.
- If you add DTL_CI columns to extraction
maps, any Insert or Delete operations result in
Null values in these columns.
Informatica recommends that you specify BA so
that you have the flexibility to use either AI or BA
for the PowerCenter Image Type connection
attribute for extraction processing.
CAPTURE_NODE The node name that the PowerExchange Logger A node name that is
uses to retrieve capture registrations and change specified in a NODE
data. statement in the
Specify this parameter only if you are using CDC dbmover.cfg file on the
offload processing with the PowerExchange local machine where the
Logger. Enter the node name of the remote PowerExchange Logger
node, as specified in a NODE statement in the runs.
dbmover.cfg file on the local machine where the
PowerExchange Logger runs. The
PowerExchange Logger uses the specified node
name to connect to the PowerExchange Listener
on the remote node to read capture registrations
and change data. The PowerExchange Logger
writes the change data to its local log files.
This parameter is optional. Default is local. Do
not specify this parameter if the capture
registrations and change data are on the local
machine where the PowerExchange Logger runs.
You can also specify an optional user ID and
password to control connection to the specified
node. For more information, see the
CAPTURE_NODE_UID parameter and the
CAPTURE_NODE_EPWD or
CAPTURE_NODE_PWD parameter.
Configuring the PowerExchange Logger 29

CAPTURE_NODE_EPWD An encrypted password that is associated with

the user ID specified in the
CAPTURE_NODE_UID parameter. This
password, in conjunction with the
CAPTURE_NODE_UID value, is used to control
PowerExchange access to capture registrations
and change data.
Tip: You can create an encrypted password in
the PowerExchange Navigator by selecting File
> Encrypt Password .
This parameter is optional. However, if you
specify CAPTURE_NODE_UID, you must enter a
password or encrypted password with either the
CAPTURE_NODE_PWD or
CAPTURE_NODE_EPWD parameter.
If you specify this parameter, do not also specify
CAPTURE_NODE_PWD.
CAPTURE_NODE_PWD A clear text password that is associated with the

user ID specified in the CAPTURE_NODE_UID
parameter. This password, in conjunction with
the CAPTURE_NODE_UID value, is used to
control PowerExchange access to capture
registrations and change data.
This parameter is optional. However, if you
specify CAPTURE_NODE_UID, you must enter a
password or encrypted password with either the
CAPTURE_NODE_PWD or
CAPTURE_NODE_EPWD parameter.
If you specify this parameter, do not also specify
CAPTURE_NODE_EPWD.

CAPTURE_NODE_UID User ID that is used to control access to capture

registrations and change data on the local
machine or on the remote node that is specified
in the CAPTURE_NODE parameter.
Whether this parameter is required depends on
the operating system of the local or remote node
and the SECURITY setting in its DBMOVER
configuration file.
If CAPTURE_NODE specifies a z/OS or i5/OS
node that has a SECURITY setting of 0, do not
specify this parameter. PowerExchange uses the
user ID under which the PowerExchange
Listener job runs to control access to capture
registrations and change data.
node that has a SECURITY setting of 1, you
must enter a valid operating system user ID for
this parameter. Otherwise, error message
PWX-00231 is issued, indicating a signon failure.
However, PowerExchange uses the user ID
under which the PowerExchange Listener job
runs to control access to capture registrations
and change data.
node that has a SECURITY setting of 2, you
must enter a valid operating system user ID for
this parameter. Otherwise, error message
PWX-00231 is issued, indicating a signon failure.
PowerExchange uses this user ID to control
access to capture registrations and change data.
If the specified user ID does not have the
authority that is required to read capture
registrations or change data, access fails.
For a Linux, UNIX, or Windows local or remote
node, enter a user ID that is valid for your data
source type:
- For DB2 for Linux, UNIX, or Windows sources,
enter a valid operating system user ID that
has DB2 DBADM or SYSADM authority.
- For Oracle sources, enter a database user ID
that permits access to Oracle redo logs and
Oracle LogMiner.
- For Microsoft SQL Server instances that use
SQL Server Authentication, enter a database
user ID that permits access to the SQL Server
distribution database. For SQL Server
instances that use Windows Authentication,
PowerExchange uses the user ID under which
the PowerExchange Listener was started. In
this case, do not specify this parameter unless
you want to specify another user.

CHKPT_BASENAME Required. An existing directory path and base Maximum length is 256.
file name that PowerExchange uses to create
checkpoint files. Checkpoint files store
information for properly resuming
PowerExchange Logger processing after a warm
start.
For example:
/capture/logger.chkpt
When creating the full checkpoint file name,
PowerExchange appends Vn, where n is a
number from 0 to (CHKPT_NUM value - 1).
For example:
/capture/logger.chkptV1.ckp
CHKPT_NUM Recommended. Number of checkpoint files to A number from 2 through

use. The PowerExchange Logger requires at 999999.
least two checkpoint files. Default is 3.
If you decrease the number of checkpoint files
after running the PowerExchange Logger, you
must cold start the PowerExchange Logger. If
you perform a warm start, the PowerExchange
Logger might restart from an incorrect location in
its log files.
COLL_END_LOG Required. PowerExchange Logger operational 0 for continuous mode.

mode. 1 for batch mode.
Options are: Default is 0.
- 0. Runs the PowerExchange Logger
continuously until you manually stop it. After
the Writer subtask completes a processing
cycle, it waits for the number of minutes
specified in the NO_DATA_WAIT parameter
before starting another processing cycle.
- 1. Runs the PowerExchange Logger in batch
mode. The PowerExchange Logger shuts
down after the seconds specified in the
NO_DATA_WAIT2 parameter elapse and no
data has been received.
COND_CDCT_RET_P Recommended. Retention period, in days, for Any number greater than 0.
CDCT records and PowerExchange Logger log Default is 60.
files. Log files that are older than this period and
their corresponding CDCT records are deleted
automatically during PowerExchange Logger
cleanup processing. Cleanup processing occurs
during startup, file switch, or shutdown
processing.
Tip: Set this parameter to minimize the size of
the CDCT file while preserving the log files that
contain the earliest change data you might need
to access. If you use continuous extraction
mode, PowerExchange reads the CDCT file each
time the interval specified in the FILEWAIT
parameter of the CAPX CAPI_CONNECTION
statement elapses. If a CDCT file becomes
large, this read activity can increase I/O, system
resource use, and latency of change data
extraction. If you use batch extraction mode, this
high read activity is not a consideration.

CONDENSENAME Optional. A name for the command-handling Maximum length is 64

service for a PowerExchange Logger for Linux, characters.
UNIX, and Windows process to which pwxcmd No default.
commands will be issued.
Syntax is:
CONDENSENAME=service_name
This service name must match the service name
that is specified in the associated SVCNODE
statement in the dbmover.cfg file. The
SVCNODE statement specifies the TCP/IP port
on which this service listens for pwxcmd
commands.
Tip: If you run the PowerExchange Logger as a
background process in continuous mode, specify
this parameter so that you can use the pwxcmd
program to issue commands to the
PowerExchange Logger. Without the use of
pwxcmd, you cannot shut down a
PowerExchange Logger process that is running
in the background or send status information to a
computer that is remote from where the
PowerExchange Logger runs.
CONDENSE_SHUTDOWN_TIMEOUT Maximum amount of time, in seconds, that the A number from 0 through
PowerExchange Logger waits after receiving the 2147483647.
SHUTDOWN or pwxcmd shutdown command Default is 600.
before stopping. During a shutdown, the
PowerExchange Logger updates the CDCT file
for each capture registration that is used to
capture change data. If you have a large number
of capture registrations, you might need to
increase this timeout period.
CONN_OVR Recommended. Name of the override Valid CAPI_CONNECTION

CAPI_CONNECTION statement to use for the name for the source type.
PowerExchange Logger. If you do not specify
CONN_OVR, the PowerExchange Logger uses
the default CAPI_CONNECTION if one is
specified in dbmover.cfg.
Informatica recommends that you specify
CONN_OVR. It is the only type of override that
the PowerExchange Logger can use.

DBID Required. A source identifier, sometimes called - For DB2 for Linux,
the instance name, that is defined in capture UNIX, and Windows,
registrations. When used with DB_TYPE, it this value is the
defines selection criteria for capture registrations Database name that is
in the CCT file. displayed for the
This value must match the instance or database registration group in the
name that is displayed in the Resource Inspector Resource Inspector.
of the PowerExchange Navigator for the - For Microsoft SQL
registration group that contains the capture Server, this value is the
registrations. Instance name that is
For Microsoft SQL Server, an instance name is displayed for the
generated when you create a registration group. registration group in the
Open the registration group in the Resource Inspector.
PowerExchange Navigator to view this Instance - For Oracle, this value is
value. the Instance name that
is displayed for the
registration group and is
also the first positional
parameter in the
ORACLEID statement in
dbmover.cfg.
If you use CDC offload
processing with the
PowerExchange Logger to
capture change data from
z/OS or i5/OS data
sources, see “Configuring
PowerExchange to
Capture Change Data on a
Remote System” on page
162 for information about
what to enter for this
parameter.
DB_TYPE Required. Source RDBMS type. - UDB for DB2 for Linux,
UNIX, and Windows
- MSS for Microsoft SQL
Server
- ORA for Oracle
If you use CDC offload
processing with the
PowerExchange Logger to
capture change data from
z/OS or i5/OS data
sources, see “Configuring
PowerExchange to
Capture Change Data on a
Remote System” on page
162 for information about
what to enter for this
parameter.
EPWD A deprecated parameter. Use

CAPTURE_NODE_EPWD instead. If both
CAPTURE_NODE_EPWD and EPWD are
specified, CAPTURE_NODE_EPWD takes
precedence.

EXT_CAPT_MASK Required. An existing directory path and a Maximum length is 256

unique prefix to be used for generating the characters.
PowerExchange Logger log files. No default.
For example:
/capture/pwxlog
Verify that no existing files match this path and
prefix. PowerExchange considers any file that
matches this path and prefix to be a
PowerExchange Logger log file, even if it is
unrelated to PowerExchange Logger processing.
To create the log files, the PowerExchange
Logger appends the following information:
.CND.CPyymmdd.Thhmmssnnn
Where:
- yymmdd is a date composed of a two-digit
year, a month, and a day.
- hhmmss is 24-hour time value, including
hours, minutes, seconds.
- nnn is a generated sequence number, which
starts from 001.
For example:
/capture/pwxlog.CND.CP080718.T1545001
Warning: Do not use the same
EXT_CAPT_MASK value for multiple
PowerExchange Logger processes. Otherwise, a
PowerExchange Logger process might corrupt
log files that are used by another
PowerExchange Logger process. Also, do not re-
use an EXT_CAPT_MASK value until the
PowerExchange Logger process has completed
processing all of the log files that match the
mask.

FILE_FLUSH_VAL Recommended. File flush interval in seconds. -1 or any number from 0

This parameter affects the latency of change through 86400.
data extractions that run in continuous extraction Default is -1.
mode. The PowerExchange Logger waits for this
interval to elapse before flushing, or writing, data
to the current log file on disk. Flushing data to
disk enables the data to be read by extractions
running in continuous extraction mode.
Valid values are:
- A -1 causes the PowerExchange Logger
process to not flush data to the current log
file. Specify this value only if you use batch
extraction mode. Do not specify this value if
you use continuous extraction mode.
Otherwise, the latency of your continuous-
mode extractions increases.
- A 0 results in a flush after every record.
- Any value from 1 through 86400 sets the flush
interval to that specific value.
Warning: A value of 0 can degrade
PowerExchange Logger and file system
performance.
Set this value as appropriate for your CDC
environment. Values that are too high can
increase change extraction latency, and values
that are too low can degrade PowerExchange
Logger and system performance. Informatica
recommends that you set this parameter to a
value that is equal to or greater than the
NO_DATA_WAIT2 value because file flushes
cannot occur until the NO_DATA_WAIT2 period
expires.
FILE_SWITCH_CRIT Type of units to use for the FILE_SWITCH_MIN - M for minutes.

and FILE_SWITCH_VAL parameters, which - R for records.
determine when to do an automatic file switch. Default is M.

FILE_SWITCH_MIN File-switch criteria that the PowerExchange - min_val. A value from -1

Logger uses when it encounters change data for through 2147483647.
a new source. You can use this parameter to - min_val_ign. A value
reduce change data latency when running from 0 through
extractions in continuous extraction mode. 2147483647.
Syntax is: Default is (-1,0).
FILE_SWITCH_MIN=(min_val,min_val_ign)
Where:
min_val is the minimum number of
FILE_SWITCH_CRIT units that must elapse after
the PowerExchange Logger encounters a
change record for a source that has no entry in
the CDCT file, before a file switch can be
performed. Valid values are:
- A -1 causes this parameter to be ignored. File
switch processing is controlled by
FILE_SWITCH_VAL only.
- A 0 causes the PowerExchange Logger to
perform a file switch each time a new source
is encountered.
- Any value from 1 through 2147483647 causes
the PowerExchange Logger to perform a file
switch when this specified number of
FILE_SWITCH_CRIT units is reached.
min_val_ign is the minimum number of
FILE_SWITCH_CRIT units that must pass during
a PowerExchange Logger cold start before the
PowerExchange Logger uses the min_val value.
Before the min_val_ign threshold is met, only
FILE_SWITCH_VAL controls file switch activity.
Valid values are:
- A 0 causes the PowerExchange Logger to use
the minimum file switch value specified in
min_val immediately after it is cold started.
- Any value from 1 through 2147483647 causes
the PowerExchange Logger to ignore the
min_val keyword for the specified number of
units.
The min_val_ign value is ignored if the
PowerExchange Logger is warm started.
Warning: The value (0,0) can result in a large
number of file switches when the
PowerExchange Logger is cold started. This
situation occurs because the PowerExchange
Logger does a file switch each time it encounters
a data source without an entry in the CDCT file.
During a cold start, the CDCT file is emptied.
Thereafter, a file switch occurs each time the
PowerExchange Logger encounters a change
record for a registered data source for the first
time.

FILE_SWITCH_VAL Number of minutes or change records, as Any number greater than 0.

determined by FILE_SWITCH_CRIT, that must Default is 30.
elapse before PowerExchange performs a file
switch.
For example, if this value is 30 and
FILE_SWITCH_CRIT=R, the PowerExchange
Logger performs a file switch every 30 records.
Or if FILE_SWITCH_CRIT=M, the
PowerExchange Logger performs a file switch
every 30 minutes.
If the PowerExchange Logger log files contain no
data when the FILE_SWITCH_VAL threshold is
reached, the file switch does not occur.
This value affects the size of the
PowerExchange Logger log files. Specify a value
that results in log files of the appropriate size for
your environment.
Tip: When using continuous extraction mode, set
this parameter such that you have larger log files
and a smaller CDCT file. When using batch
extraction mode, set this parameter to a value
that causes file switches to occur within the
timeframe that meets your change extraction
latency requirements.
GROUPDEFS Path and file name of the optional Maximum length is 255
PowerExchange Logger group definition file. characters.
This file defines groups of capture registrations No default.
that the PowerExchange Logger uses to capture
change data to separate sets of log files. It also
defines the path that the PowerExchange Logger
uses to create the log files that contain the
change data for each group. This parameter is
optional.
LOGGER_DELETES_EXPIRED_CDCT_RE Controls how expired CDCT records, for which Y or N

CORDS the retention period has elapsed, are deleted. Default is Y.
Options are:
- Y. The PowerExchange Logger maintains the
CDCT retention array and deletes expired
CDCT records during file switches. If you
enter Y, you cannot issue the
DELETE_EXPIRED_CDCT command from the
PWXUCDCT utility to delete expired CDCT
records.
- N. The PowerExchange Logger does not
maintain the CDCT retention array and does
not delete expired CDCT records. However,
you can issue the DELETE_EXPIRED_CDCT
command from the PWXUCDCT utility to
delete expired CDCT records. To use the
DELETE_EXPIRED_CDCT command, you
must specify N.
Note: This parameter does not affect
PowerExchange Logger deletions of CDCT
records rolled back because of a cold start or
a warm start to a prior point in time.

MAX_RETENTION_EXPIRY_DAYS Maximum number of days to hold retention array A number from 1 through
items in memory. Retention array items define 999.
when CDCT records expire and indicate the log Default is 999.
file names and registration tags referenced by
the CDCT records.
Tip: If you have a large volume of CDCT
records, you can usually avoid memory
shortages by setting the
LOGGER_DELETES_EXPIRED_CDCT_RECOR
DS parameter to N and running the PWXUCDCT
utility DELETE_EXPIRED_CDCT command on a
regular, scheduled basis. Use the
MAX_RETENTION_EXPIRY_DAYS parameter
only in situations with extreme memory
limitations.
NO_DATA_WAIT If you run the PowerExchange Logger in 0 or greater.

continuous mode, specify the number of minutes Default is 60.
that the PowerExchange Logger must wait
before starting the next logging cycle. A value of
0 causes no waiting to occur between
PowerExchange Logger processing cycles. If
source data is not available, the CAPI sleeps.
For continuous extraction mode, this value
should be low so that the next logging cycle
starts shortly after the current one completes.
If the value of FILE_SWITCH_CRIT is M and the
value of FILE_SWITCH_VAL is less than the
value of NO_DATA_WAIT, the PowerExchange
Logger uses the FILE_SWITCH_VAL value
instead.
NO_DATA_WAIT2 Number of seconds that PowerExchange waits at Any number greater than 0.
the end-of-log for more change data before Recommended value is 2.
returning control to the PowerExchange Logger. Default is 600.
If this wait period elapses and new change data
has not been received, PowerExchange returns
control to the PowerExchange Logger, and the
PowerExchange Logger then stops the current
logging cycle.
The recommended value is 2. If you enter a
higher value, execution of commands for the
PowerExchange Logger might be delayed.

PROMPT When you run the PowerExchange Logger in Y or N

foreground mode, controls whether Default is Y for a
PowerExchange displays a user confirmation PowerExchange Logger
prompt and waits for a response when you that runs in foreground
perform one of the following actions: mode.
- Cold start the PowerExchange Logger. Default is N for a
- Warm start the PowerExchange Logger from a PowerExchange Logger
previous position in the change stream. This process that runs in
situation occurs only if checkpoint files that background mode or as a
were more recent than the current ones were PowerExchange Logger
deleted, and the CDCT file still contains Service in the Informatica
records related to the deleted files. domain.
Options are:
- Y. Displays the confirmation message
PWX-33236 for a cold start or PWX-33242 for
a warm start. You must respond to the
message for startup processing to continue.
- N. Does not display the confirmation
messages. PowerExchange attempts to start
without first prompting for user confirmation.
If you run the PowerExchange Logger in
foreground mode, the default is Y.
If you run the PowerExchange Logger in
background mode or as a PowerExchange
Logger Service in the Informatica domain, the
default is N. In this case, if you enter
PROMPT=Y in the pwxccl.cfg file, the
PowerExchange Logger ignores this setting,
issues error message PWX-33253, and
continues processing.
PWD A deprecated parameter. Use

CAPTURE_NODE_PWD instead. If both
CAPTURE_NODE_PWD and PWD are specified,
CAPTURE_NODE_PWD takes precedence.

RESTART_TOKEN and Parameters that define a restart point for starting - Specific restart and
SEQUENCE_TOKEN change data processing when a PowerExchange sequence token values.
Logger is cold started. - 0
A restart point is defined by both a restart token - Not specified.
and a sequence token. If you use CDC offload
Depending on how you set these parameters, processing with the
PowerExchange Logger processing starts from PowerExchange Logger to
one of the following restart points during a cold capture change data from
start. z/OS or i5/OS data
If you do not specify these parameters, sources, see the
processing starts from the current end-of-log PowerExchange Condense
position. chapter in the
If you enter 0 for both parameters, processing PowerExchange CDC
starts from the default start location: Guide for i5/OS and
- For DB2, the default location is the current log PowerExchange CDC
position at the time the PowerExchange Guide for z/OS for
capture catalog was created. information about what to
- For Oracle, the default location is the most enter for these parameters.
current Oracle catalog dump.
- For Microsoft SQL Server, the default location
is the oldest data available in the publication
database.
If you enter restart token and sequence token
values other than 0, processing resumes from
the specific restart point defined by these token
values.
SIGNALLING Indicates whether the PowerExchange Logger Y or N

attempts to take automatic action in the event of Default is N.
certain errors.
Options are:
- N. The PowerExchange Logger does not
automatically trap and handle system errors.
Instead, the operating system uses default
error handling. Usually, the default handing is
to report the program line in error and dump
memory.
- Y. The PowerExchange Logger automatically
handles certain errors such as memory
corruption. After the PowerExchange Logger
handles the error, it attempts to shut down in
a controlled manner.

UID A deprecated parameter. Use

CAPTURE_NODE_UID instead. If both
CAPTURE_NODE_UID and UID are specified,
CAPTURE_NODE_UID takes precedence.
VERBOSE Indicates whether the PowerExchange Logger - Y for verbose messaging

writes verbose or terse messages to the - N for terse messaging
PowerExchange message log file for activities Default is Y.
that it performs frequently, such as cleanup,
checkpoint, condense, and file-switch processing.
Options are:
- N. The PowerExchange Logger logs a single
terse message for each file switch and
checkpoint.
- Y. The PowerExchange Logger logs multiple
messages at various processing points, such
as when starting or ending a cycle of reading
source data or doing a file switch. Verbose
messaging often includes processing statistics
such as records processed and elapsed time.
RELATED TOPICS:
¨ “PowerExchange Logger Operational Modes” on page 25
¨ “Configuring PowerExchange to Capture Change Data on a Remote System” on page 162
Example pwxccl.cfg
PowerExchange provides an example pwxccl.cfg file in the PowerExchange installation directory, which you can
customize.
The example file contains the following statements:

/* Name for PWXCMD control
/*CONDENSENAME=PWXCCL1
DBID=ORACOLL1
DB_TYPE=ORA
CAPTURE_NODE_UID=user_id
CAPTURE_NODE_EPWD=encrypted_password
/* CAPTURE_NODE_PWD=plain_text_password
PROMPT=Y
EXT_CAPT_MASK=/capture/condenseO
CHKPT_NUM=3
CHKPT_BASENAME=/capture/condenseO.chkpt
COND_CDCT_RET_P=50
LOGGER_DELETES_EXPIRED_CDCT_RECORDS=Y
/* 0 = continuous, 1 = Stop at end-of-log (batch)

COLL_END_LOG=0
/* Number of minutes to wait between CAPI read cycles in seconds

NO_DATA_WAIT=0
/* Number of seconds to wait at the end-of-log for more change data
NO_DATA_WAIT2=60
/* Number of seconds before flushing, or writing, data to the current log file on disk
/* -1 = No flush, 0 = flush every record, 1 to N flush every N seconds
/*FILE_FLUSH_VAL=60
/* Minimum number of FILE_SWITCH_CRIT units after new CDCT source entry (normal,coldstart)
/*FILE_SWITCH_MIN=(0,0)

FILE_SWITCH_CRIT=M
FILE_SWITCH_VAL=20
CAPT_IMAGE=BA
SEQUENCE_TOKEN=00
RESTART_TOKEN=00
Note: If you enter values in the EXT_CAP_MASK and CHKPT_BASENAME parameters that include spaces, you
must enclose the values in double quotation marks.
Customizing dbmover.cfg for the PowerExchange Logger

To use the PowerExchange Logger, you must define the CAPT_PATH statement and certain source-specific
statements in the dbmover.cfg file.
Also, you can include some optional parameters to help make finding messages for the PowerExchange Logger
easier or to send commands to a PowerExchange Logger process that is running in background mode.
Use the following key parameters:
CAPT_PATH
Required. Specifies the path to the directory where the CCT and CDCT files reside. The CCT file contains
information about capture registrations. The CDCT file contains information about the PowerExchange Logger
log files, such as file names and number of records.
LOGPATH
Optional. Specifies a unique path to the PowerExchange message log files. Use this parameter to create
message log files in a directory that is separate from your current working directory so that you can find the
message log files more easily.
SVCNODE
Optional. Specifies the TCP/IP port on which a command-handling service for a PowerExchange Logger
process listens for commands that you issue with the pwxcmd program. You must define this parameter if you
run the PowerExchange Logger process in background mode on a Linux or UNIX system. For more
information about pwxcmd commands, see the PowerExchange Command Reference.
TRACING
Optional. Enables alternative logging. PowerExchange creates a set of alternative log files for each
PowerExchange process in a separate directory. You can specify the directory location, the number of log
files, and the log file size in MB. When a log file reaches the specified size, PowerExchange switches to the
next log file and begins overwriting any data in that file. Alternative logging is faster and enables you to
customize the amount of data logged for long-running jobs, such as a PowerExchange Logger process that
runs in continuous mode. If you specify this statement, also specify the LOGPATH statement.
In addition to these parameters, the PowerExchange Logger requires source-specific statements, such as the
ORCL CAPI_CONNECTION, UOWC CAPI_CONNECTION, and ORACLEID statements for Oracle.
For more information about all DBMOVER configuration parameters, see the PowerExchange Reference Manual.
RELATED TOPICS:

Using PowerExchange Logger Group Definitions
To create separate sets of PowerExchange Logger log files for groups of tables, create a PowerExchange Logger
group definition file. Then, specify its path and file name in the GROUPDEFS parameter of the pwxccl.cfg file.
When the PowerExchange Logger process starts, it reads the group definition file and creates a separate set of
log files for each defined group.
Group definitions can help improve the efficiency of extraction sessions because the extractions target a more
specific set of PowerExchange Logger log files.
By default, the PowerExchange Logger processes change data for all tables that reside on the instance specified
by the DBID parameter and that have active capture registrations with the Condense option set to Part. Changes
for all of these tables are written to a single set of log files (not taking into account file switching). For a table with
a low level of change activity, the PowerExchange Logger might need to read many change records in the log files
before finding the changes of interest.
With group definitions, you can define a group that includes a subset of capture registrations. The PowerExchange
Logger then writes change data to a separate set of log files for the tables that are associated with these
registrations. When an extraction process runs, it is more likely to find the change data for a table in the group
faster because it reads only the log files for that group.
For example, if you have five source tables with a low level of change activity and one table with a high level of
change activity, you can define a group that includes the low-activity tables and another group that includes only
the high-activity table only. Then, in PowerCenter, define a CDC session that extracts change data from the
PowerExchange Logger log files for the low-activity group, and define another CDC session that extracts change
from the log files for the high-activity group. This configuration enables the CDC session for the low-activity tables
to find and extract the few change records for these tables much more quickly.
If you have multiple tables with the same table name but different schemas, you can define a single capture
registration for the table and specify it once, under a single group, in the group definition file. For any other group
that includes the same table with a different schema, you can override the schema name in the group definition by
using a SCHEMA statement. By using the SCHEMA statement, you can avoid creating multiple capture
registrations and specifying each one in the group definition file. For example, if you have an EMPLOYEE table
with different schemas for the north, south, east, and west regions, you can register the north EMPLOYEE table
only and specify the capture registration name in the NORTH group. Then specify only the override schemas in
the EAST, WEST, and SOUTH groups.
Note: SCHEMA statements are optional for DB2 for i5/OS sources and for DB2 and Oracle sources on Linux,
UNIX, and Windows. SCHEMA statements are not supported for SQL Server sources on Windows or any data
source on z/OS.
On Linux, UNIX, and Windows, PowerExchange requirements for unregistered versions of tables, for which a REG
statement is not specified, vary by source type:
¨ For DB2 for Linux, UNIX, and Windows, you must define any unregistered version of a table with the DATA
CAPTURE CHANGES clause.
¨ For Oracle, you must create an Oracle supplemental log group for the unregistered table, which is similar to the
supplemental log group that was created for the registered copy of the table at registration completion.
¨ For Microsoft SQL Server, you must register all versions of a table in PowerExchange and specify a REG
statement in the group definition file.
Tip: When using group definitions, you can optimize extraction efficiency by defining a CDC session in
PowerCenter for each group of tables defined in the group definition file.
RELATED TOPICS:
¨ “PowerExchange Logger Group Definition File” on page 45

PowerExchange Logger Group Definition File
A PowerExchange Logger group definition file contains one or more GROUP statements. Each GROUP statement
contains REG or SCHEMA parameters that directly or indirectly identify a group of capture registrations and tables
for which you want to create separate sets of PowerExchange Logger log files.
For the PowerExchange Logger to use the group definition file, you must specify the path and file name of the file
in the GROUPDEFS parameter of the pwxccl.cfg file.
Note: If you specify the GROUPDEFS parameter, the PowerExchange Logger ignores the EXT_CAPT_MASK
parameter in the pwxccl.cfg file when creating log files.
The following table describes the statements and parameters in the group definition file:
Statement Positional Parameter Description Data Type and

Length
GROUP group_name A unique user-defined name for the group. This VARCHAR(255)
parameter is required.
external_capture_mask A unique path and file-name prefix for the VARCHAR(255)

PowerExchange Logger log files that are created
for tables in the group. This parameter is required.
Note: This path and prefix is used for the group
instead of any the path and prefix that are
specified in the EXT_CAPT_MASK parameter of
the pwxccl.cfg file.
REG registration_name Optional. Registration name that is specified in the VARCHAR(8)

Name field of a capture registration. This
lowercase name can be the full registration name
or the first part of the name followed by an
asterisk (*) wildcard. This parameter is optional. If
omitted, the PowerExchange Logger assumes
REG=*.
SCHEMA schema_name Optional. Name of the override schema. You can VARCHAR(255)
optionally use this parameter for DB2 for i5/OS
sources and for DB2 and Oracle sources on Linux,
UNIX, and Windows.
Note: This parameter is not supported for SQL
Server sources on Windows. If you use the
offloading feature to have the PowerExchange
Logger process data from z/OS sources, this
parameter is also not supported for the z/OS
sources.
Use the following rules and guidelines when you create a PowerExchange Logger group definition file:
¨ Each group_name must be unique within the group definition file.
¨ Each external_capture_mask must be unique on the system.
¨ SCHEMA statements are optional for DB2 for i5/OS sources and for DB2 and Oracle sources on Linux, UNIX,
and Windows. SCHEMA statements are not supported for SQL Server sources on Windows or any data source
on z/OS.
¨ If you use a SCHEMA statement, you must define a capture registration in the group. You can specify multiple
SCHEMA statements under a GROUP if you want the tables with those schemas to be included in the group.
¨ REG statements apply to the preceding SCHEMA statement. If a SCHEMA statement is not present, the REG
statements apply to the preceding GROUP statement.

¨ If the file contains a SCHEMA or REG statement without a preceding GROUP statement, the PowerExchange
Logger issues a syntax error.
¨ Do not include the same schema.table value in more than one group. If a table is included in multiple groups,
only the first group that includes the table logs changes for it.
¨ If you do not define at least one REG statement for a GROUP, the PowerExchange Logger includes all of the
active capture registrations that are defined for the specified DBID instance and for which the Condense option
is set to Part.
¨ If a registration belongs to multiple groups, the PowerExchange Logger logs changes for that registration only
under the first group in the group definition file that includes the registration.
Example Group Definition File

PowerExchange provides an example group definition file, pwxcclgrp.cfg, in the PowerExchange installation
directory. Use this example as a starting point when creating your group definition file.
The example file contains the following statements:

GROUP=(Company1People,"/user/logger_files/people/company1/condense")
REG=Emp*
REG=Manager
GROUP=(UK_People,"/user/logger_files/people/UK/condense")
SCHEMA=Company2
REG=Manager
REG=Emp*
REG=Em*
SCHEMA=Company3
REG=Manager
REG=Emp*
GROUP=(All_Managers,"/user/logger_files/people/managers/condense")
SCHEMA=Company1
REG=Manager
SCHEMA=Company2
REG=Manager
SCHEMA=Company3
REG=Manager
GROUP=(AllCompany3_Locations,"/user/logger_files/locations/company3/condense")
REG=loc*
GROUP=(Company2Jobs,"/user/logger_files/jobs/company2/condense")
REG=Job*
Note: Because this example is for a group definition file on a Linux or UNIX system, the paths include forward
slashes. A group definition file on Windows system would be similar but have back slashes.
This example file defines the following groups:
¨ Company1People group. Groups all tables associated with capture registrations that have names beginning
with “Emp” or the name “Manager.” Changes for these tables are logged to log files that have file names
beginning with “condense” and that are located at “/user/logger_files/people/company1/.”
¨ UK_People group. Groups tables that have the schema Company2 and that are associated with capture
registrations that have names beginning with “Emp” or “Em” or the name “Manager.” Changes for these tables
are logged to log files that have names beginning with “condense” and that are located at “/user/logger_files/
people/UK/.”
¨ All_Managers group. Groups tables that have the schema Company1, Company2, or Company3 and that are
associated with the capture registration with the name “Manager.” Changes for these tables are logged to log
files that have names beginning with “condense” and that are located at “/user/logger_files/people/managers/.”

¨ AllCompany3_Locations group. Groups all tables that are associated with capture registrations that have
names beginning with “loc.” Changes for these tables are logged to log files that have names beginning with
“condense” and that are located at “/user/logger_files/locations/company3/.”
¨ Company2Jobs group. Groups all tables that are associated with capture registrations that have names
beginning with “Job.” Changes for these tables are logged to log files that have names beginning with
“condense” and that are located at “/user/logger_files/jobs/company2/.”
Some tables might be included in more than one group. For example, the table COMPANY2.MANAGERS is in the
Company1People, UK_People, and All_Managers groups. However, changes for this table are logged only under
the Company1People group because it is the first group in the file that includes this table.
Starting the PowerExchange Logger

You can cold start or warm start the PowerExchange Logger process.
¨ A cold start uses the restart and sequence tokens, if present, in the pwxccl.cfg configuration file to determine
the point in the change stream from which the PowerExchange Logger starts reading changes. If you are
starting the PowerExchange Logger for the first time, you must perform a cold start.
¨ A warm start uses the restart and sequence tokens in the last checkpoint file to resume CDC processing. You
can perform a warm start only if you have run the PowerExchange Logger previously and have recent
checkpoint files.
You cannot use the pwxcmd program to start the PowerExchange Logger.
PWXCCL Syntax and Parameters

To start the PowerExchange Logger process, run the pwxccl program, which is located in the PowerExchange
installation directory by default.
PWXCCL Syntax
The pwxccl statement has the following syntax:
pwxccl
[coldstart={Y|N}]
[config=path/pwx_config_file]
[cs=path/pwxlogger_config_file]
[license=path/license_file]
Use the following rules and guidelines when you enter the pwxccl statement:
¨ To cold start the PowerExchange Logger, you must set the coldstart parameter to Y. The default is N.
¨ All parameters are optional. However, if you specify the config or license parameter, the cs parameter is
required.
¨ In the config, cs, and license parameters, the full path is required only if the file is not in the default location.
¨ On Linux and UNIX, append an ampersand (&) at the end of the statement to run the PowerExchange Logger
in background mode.
For more information about pwxccl syntax, see the PowerExchange Command Reference.
PWXCCL Parameters
You can specify several optional parameters in the pwxccl statement.
Starting the PowerExchange Logger 47

The following table describes each parameter:
Parameter Description
coldstart Indicates whether to cold start or warm start the PowerExchange Logger.
Enter one of the following values:
- Y. Cold starts the PowerExchange Logger. You must specify COLDSTART=Y to perform a cold start.
The absence of checkpoint files does not trigger a cold start. If you specify Y and checkpoint files exist,
the PowerExchange Logger ignores the files. If the CDCT file contains records, the PowerExchange
Logger deletes these records.
- N. Warm starts the PowerExchange Logger from the restart point that is indicated in the last checkpoint
file. If no checkpoint file exists in the CHKPT_BASENAME directory, the PowerExchange Logger ends
with error message PWX-33227.
Default is N.
config Full path and file name for a DBMOVER configuration file that overrides the default dbmover.cfg file in the
installation directory. The override file must have a path or file name that is different from that of the default
file.
This override file takes precedence over any other override configuration file that you optionally specify
with the PWX_CONFIG environment variable.
cs Full path and file name of the PowerExchange Logger configuration file. Use this parameter to specify a
PowerExchange Logger configuration file that overrides the default pwxccl.cfg in the installation directory.
The override file must have a path or file name that is different from that of the default file.
license Full path and file name for a license key file that overrides the default license.key file in the installation
directory. The override file must have a file name or path that is different from that of the default file.
This override file takes precedence over any other override license key file that you optionally specify with
the PWX_LICENSE environment variable.
Note: In these parameters, the full path is required only if the file is not in the default location.
How the PowerExchange Logger Determines the Start Point for a

Cold Start
When you cold start a PowerExchange Logger process, it uses the RESTART_TOKEN and SEQUENCE_TOKEN
parameters, if present, in the pwxccl.cfg configuration file to determine the point in the change stream at which to
start reading changes.
Based on how you set these parameters, the PowerExchange Logger starts from one of the following points in the
change stream:
¨ If you do not define the RESTART_TOKEN and SEQUENCE_TOKEN parameters, the PowerExchange
Loggers starts from the current end-of-log (EOL), or current point in time in the change stream.
Tip: You can generate restart and sequence tokens for the current EOL by running the DTLUAPPL utility with
the RSTTKN GENERATE parameter or by performing a database row test with the SELECT
CURRENT_RESTART SQL statement in PowerExchange Navigator.
¨ If you enter only zeroes (a single 0, or an even number of 0s) in the RESTART_TOKEN and
SEQUENCE_TOKEN parameters, the PowerExchange Logger starts from the oldest available change record in
the change stream.
¨ If you enter valid restart information in the RESTART_TOKEN and SEQUENCE_TOKEN parameters, the
PowerExchange Logger starts from the point in the change stream that the token values identify. Use this
method to start the PowerExchange Logger from a specific point.

Cold Starting the PowerExchange Logger
Use this procedure to cold start the PowerExchange Logger. In the start statement, you must include
COLDSTART=Y.
During a cold start, the PowerExchange Logger ignores any checkpoint files that exist in the directory that is
specified by the CHKPT_BASENAME parameter in the pwxccl.cfg file. If the CDCT file contains records, the
PowerExchange Logger deletes these records.
To cold start the PowerExchange Logger:
1. If you previously ran the PowerExchange Logger and have existing checkpoint, CDCT, and log files, retain
these files for historical purposes.
You can move or rename the files, as long as another PowerExchange Logger process is not using them. Do
delete them if you want to retain change processing history.
Warning: If you delete, move, or rename the CCT file, the capture registrations will not be available.
2. In the pwxccl.cfg configuration file, set the RESTART_TOKEN and SEQUENCE_TOKEN parameters in a
manner that causes the PowerExchange Logger to start from the appropriate point in the change stream.
3. To cold start the PowerExchange Logger, enter the following statement at the command line:
pwxccl coldstart=y
Include the optional config, cs, and license parameters if you want to override the default dbmover.cfg,
pwxccl.cfg, and license.key files. On Linux and UNIX systems, you can add an ampersand (&) at the end of
the statement to run the PowerExchange Logger in background mode. For more information about
PowerExchange Logger syntax, see the PowerExchange Command Reference.
RELATED TOPICS:
¨ “How the PowerExchange Logger Determines the Start Point for a Cold Start” on page 48
¨ “PWXCCL Parameters” on page 47
Managing the PowerExchange Logger

To assess the status of the PowerExchange Logger for Linux, UNIX, and Windows, you can display messages
about PowerExchange Logger processing, memory use, and CPU use.
Occasionally, you might need to stop the PowerExchange Logger.
Commands for Controlling and Stopping PowerExchange Logger

Processing
Use PowerExchange Logger for Linux, UNIX, and Windows commands to manually initiate a file switch or another
logging cycle, stop the PowerExchange Logger, or display messages about PowerExchange Logger processing
and system resource use.
You can enter these commands from the command line or by using the pwxcmd program. The output is displayed
on screen and written to the PowerExchange message log.
Note: To use pwxcmd, you must specify the CONDENSENAME parameter in the pwxccl.cfg file and the
SVCNODE statement in the dbmover.cfg file.
Managing the PowerExchange Logger 49

The following table describes each command:
Command-line Command pwxcmd Command Description
CONDENSE condense When the PowerExchange Logger is

running in continuous mode, manually
starts a new PowerExchange Logger
logging cycle before the wait period for
starting another cycle has elapsed. The
wait period is defined by the
NO_DATA_WAIT parameter in
pwxccl.cfg.
DISPLAY ALL displayall Displays all messages that can be

produced by the other PowerExchange
Logger DISPLAY commands, arranged
by command.
DISPLAY CHECKPOINTS displaycheckpoints Displays message PWX-26041, which

reports information about the latest
checkpoint file. The information
includes the file sequence number,
timestamp, number of data records and
commit records, and commit time.
DISPLAY CPU displaycpu Displays the CPU time spent, in

microseconds, for PowerExchange
Logger processing during the current
logging cycle, by processing phase.
Also includes the total CPU time for all
PowerExchange Logger processing.
Processing phases include reading
source data, writing data to log files,
performing file switches, and performing
“other” processing such as initialization.
DISPLAY EVENTS displayevents Displays events that the

PowerExchange Logger Controller,
Command Handler, and Writer tasks
are waiting on. Also indicates if the
Writer is processing data or is in a
sleep state waiting for an event or
timeout to occur.
DISPLAY MEMORY displaymemory Displays PowerExchange Logger

memory use, in bytes, for each
PowerExchange Logger task and
subtask, with totals for the entire
PowerExchange Logger process.
Memory use is reported for the
following categories: Application, Total,
and Maximum.

DISPLAY RECORDS displayrecords Displays counts of change records that

the PowerExchange Logger processed
during the current processing cycle. If
the PowerExchange Logger did not
receive changes during the current
cycle, displays counts of change
records for the current set of
PowerExchange Logger log files.
Record counts are shown by record
type. Record types are Delete, Insert,
Update, Commit, and Total.
DISPLAY STATUS displaystatus Displays the status of the

PowerExchange Logger Writer subtask,
for example, initializing, writing source
data to a PowerExchange Logger log
file, or starting a checkpoint.
FILESWITCH fileswitch Closes open PowerExchange Logger

log files if they contain data and then
switches to a new set of log files. If the
log files do not contain data, the file
switch does not occur.
If you use batch extraction mode, you
can use this command to make change
data in the current log files available for
extraction processing before the next
file switch is due to occur. To issue the
fileswitch command from a script or
batch file, you must use the pwxcmd
program. Usually, you do not need to
perform manual file switches if you use
continuous extraction mode.

SHUTCOND shutcond Stops the PowerExchange Logger in a

controlled manner after initiating and
completing a final logging cycle. The
final logging cycle enables the
PowerExchange Logger to capture all of
the changes up to point when the
command is issued. After the logging
cycle completes, the PowerExchange
Logger closes open log files, updates
the CDCT file, takes a final checkpoint
to record the latest restart and
sequence tokens, closes the CAPI,
stops the Writer and Command Handler
subtasks, and then ends the pwxccl
program. Use this command if a logging
cycle has not run recently.
SHUTDOWN shutdown Stops the PowerExchange Logger in a

controlled manner after closing any
open PowerExchange Logger log files
and writing the latest restart position to
the checkpoint files. During shutdown
processing, the PowerExchange Logger
closes open log files, updates the
CDCT file, takes a final checkpoint,
closes the CAPI, stops the Writer and
Command Handler subtasks, and then
ends the pwxccl program. Use this
command to stop a PowerExchange
Logger process that is running in
continuous mode.
For more information about command syntax, example output, and pwxcmd use, see the PowerExchange
Command Reference.
Assessing PowerExchange Logger Performance

To assess PowerExchange Logger performance, you can view key PowerExchange Logger messages that report
CPU use and elapsed times for processing.
Enter VERBOSE=Y in the pwxccl.cfg configuration file to have the PowerExchange Logger produce more detailed
messages during initialization, condense, fileswitch, record expiration, and shutdown processing. For example, the
following verbose messages indicate CPU use by the Writer subtask:
¨ Message PWX-33274 is issued before the Writer subtask starts reading source data after initialization and
before the PowerExchange Logger shuts down:
PWX-33274 CPU Total number. CAPI Read number. Writing number. File switching number. Other number
¨ Message PWX-33279 issued after each file switch and checkpoint:
PWX-33279 CPU total number. This file total number. CAPI Reads number. Writing file number. Other
number

If you do not use verbose messaging, you can use the DISPLAY CPU and DISPLAY RECORDS commands to
gather statistics that are useful for assessing PowerExchange Logger performance and status.
¨ The DISPLAY CPU command displays the CPU time spent, in microseconds, for PowerExchange Logger
processing during the current logging cycle, by processing phase and with the total for all processing.
Processing phases include:
- Reading source data
- Writing data to PowerExchange Logger log files
- Performing file switches
- Performing "other processing," such as initialization and Command Handler processing of commands
¨ The DISPLAY RECORDS command displays counts of change records that the PowerExchange Logger
processed during the current processing cycle. If the PowerExchange Logger did not receive changes during
the current cycle, the command displays counts of change records for the current PowerExchange Logger log
files. Record counts are shown for each type of change record processed and for total records processed.
Change record types include Delete, Insert, Update, and Commit.
For more information about these commands, including example output, see the PowerExchange Command
Reference.
Maintaining the PowerExchange Logger CDCT File and Log Files

You can use the PWXUCDCT utility to maintain the PowerExchange Logger CDCT file and log files.
Use the following utility commands to perform maintenance tasks:
Command Description
CREATE_CDCT_BACKUP Back up all CDCT records for the source instance that is
specified in the DBID parameter of the pwxccl.cfg
configuration file.
DELETE_EXPIRED_CDCT Delete CDCT records for which the retention period has
expired and any PowerExchange Logger log files that are
referenced by those records. Use this command only if you
set the LOGGER_DELETES_EXPIRED_CDCT_RECORDS
parameter to N in the pwxccl.cfg file.
DELETE_ORPHAN_FILES Delete PowerExchange Logger log files that are not

referenced by any record in the CDCT file.
DERIVE_CDCT_BACKUP Create a backup of the CDCT file based on PowerExchange

Logger log files, if the original backup file is damaged or
deleted.
REPORT_CDCT List information about the CDCT file and its records. For each
CDCT record, the command reports the record number,
registration tag name, log file name, number of change
records received for the registered table, start and end times,
and start and end restart tokens.
REPORT_CDCT_BY_TIME List CDCT records in the order in which they expire.
REPORT_CONFIG List the parameter settings in the PowerExchange Logger

pwxccl.cfg configuration file.

Command Description
REPORT_CHECKPOINTS List checkpoint files in chronological order, from earliest to

latest, based on when they were written. For each file, the list
provides information such as number of capture registrations
processed, reason for the checkpoint, sequence and restart
tokens, number of expired CDCT records that were deleted,
and number of log files to which change data was written.
REPORT_EXPIRED_CDCT List PowerExchange Logger log files based on their file

names.
REPORT_FILES_BY_TIME List PowerExchange Logger log files in the order in which

they were created, from earliest to latest.
REPORT_ORPHAN_FILES List PowerExchange Logger log files that are not referenced
by any record in the CDCT file.
RESTORE_CDCT Restore the CDCT file from a backup if the CDCT file is
damaged or deleted.
For more information about the PWXUCDCT utility, see the PowerExchange Utilities Guide.
Backing Up PowerExchange Logger Files

Periodically, back up the PowerExchange Logger for Linux, UNIX, and Windows checkpoint files, CDCT file, and
log files. If existing files become damaged or deleted, you can then use the backups to restore the files.
If possible, back up the PowerExchange Loggers files during a period when source data is not being written to the
PowerExchange Logger log files. Back up the files in the following sequence to ensure that you have a checkpoint
file that matches the backup:
1. Checkpoint files
2. CDCT file
3. PowerExchange Logger log files
Note: During a file switch, the Writer subtask processes the files in the reverse sequence.
To back up the CDCT file, you can use the PWXUCDCT utility CREATE_CDCT_BACKUP command.
Re-creating the CDCT File After a Failure

If the CDCT file and its recent backups are damaged or deleted, you can re-create the CDCT file based on the
PowerExchange Logger log files. You must derive a CDCT backup based on the current PowerExchange Logger
log files and then restore that backup.
1. Issue PWXUCDCT utility DERIVE_CDCT_BACKUP command.

For more information about using PWXUCDCT utility commands, see the PowerExchange Utilities Guide.
2. Restore the derived backup by issuing the PWXUCDCT utility RESTORE_CDCT command.
3. Verify that the restore operation was successful as follows:
¨ Verify that the return code from the PWXUCDCT utility is zero.
¨ Verify that messages PWX-25140 through PWX-25145 provide reasonable record counts for the records
read from the backup file and for the records that were changed in the CDCT file.

Part III: PowerExchange CDC Data
Sources
¨ DB2 for Linux, UNIX, and Windows Change Data Capture, 56
¨ Microsoft SQL Server Change Data Capture, 70
¨ Oracle Change Data Capture with Oracle LogMiner, 80
55
CHAPTER 4
DB2 for Linux, UNIX, and Windows

Change Data Capture
¨ DB2 for Linux, UNIX, and Windows CDC Overview, 56
¨ Planning for DB2 CDC, 57
¨ Configuring DB2 for CDC, 58
¨ Configuring PowerExchange for DB2 CDC, 59
¨ Using a DB2 Data Map, 65
¨ Managing DB2 CDC, 66
¨ DB2 for Linux, UNIX, and Windows CDC Troubleshooting, 69
DB2 for Linux, UNIX, and Windows CDC Overview

PowerExchange captures change data from the DB2 for Linux, UNIX, and Windows recovery logs for the database
that contains your source tables. PowerExchange uses the PowerExchange Client for PowerCenter (PWXPC) to
coordinate with PowerCenter to move the captured change data to one or more targets.
For PowerExchange to capture DB2 change data, you must perform the following configuration tasks in DB2:
¨ Ensure that archive logging is active for the database.
¨ Create a PowerExchange capture catalog table in the database. The capture catalog table stores information
about all tables in the source database, including column definitions and DB2 log positions.
Also, perform the following configuration tasks in PowerExchange:
¨ Define a capture registration for each source table. In the capture registration, you can select a subset of
columns for which to capture data. PowerExchange generates a corresponding extraction map. Optionally, you
can define an additional extraction map.
¨ If a source table contains columns in which you store data in a format that is inconsistent with the column
datatype, you can optionally create a data map to manipulate that data with expressions. For example, if you
store packed data in a CHAR column, you can create a data map to manipulate and prepare that data for
56
loading to a target. You must merge the data map with the extraction map for the source table during capture
registration creation.
¨ If you want to use the PowerExchange Logger for Linux, UNIX, and Windows to capture change data and write
it to PowerExchange Logger log files, configure the PowerExchange Logger. The change data is then extracted
from the PowerExchange Logger log files. Benefits of the PowerExchange Logger include fewer database
accesses, faster CDC restart, and no need to prolong retention of DB2 log files for change capture.
PowerExchange works in conjunction with PowerCenter to extract change data from DB2 recovery logs or
PowerExchange Logger log files and load that data to one or more targets.
RELATED TOPICS:
¨ “Extracting Change Data” on page 125
Planning for DB2 CDC

Before you configure DB2 for Linux, UNIX, and Windows CDC, verify that the following prerequisites and user
authority requirements are met. Also, review the restrictions so that you can properly configure CDC.
Prerequisites
PowerExchange CDC has the following prerequisites:
¨ Archive logging must be active for the database that contains the source tables from which change data is to
be captured.
¨ DB2 source tables must have been defined with the DATA CAPTURE CHANGES clause for capture processing
to occur.
Required User Authority

For PowerExchange to read change data from DB2 logs, the user ID that you specify for database access must
have SYSADM or DBADM authority. Usually, you specify this user ID in the UDB CAPI_CONNECTION statement
in the dbmover.cfg file.
Planning for DB2 CDC 57

CDC Restrictions
The following restrictions apply to DB2 CDC processing:
¨ To extract change data on a DB2 client machine that is remote from the DB2 server where the change data is
captured, both machines must have the same architecture. Otherwise, change data capture processing might
fail with the error message PWX-20628.
¨ PowerExchange cannot capture change data for the following DB2 datatypes:
- DECFLOAT, LOB, and XML datatypes. You can create a capture registration for a table that includes columns
with DECFLOAT, LOB, and XML datatypes. However, the registration does not include these columns, and
PowerExchange does not capture change data for them. PowerExchange does capture change data for the
other columns in the registered table that have supported datatypes.
- User-defined datatypes. Tables that include columns with user-defined datatypes cannot be registered for
change data capture. PowerExchange cannot capture change data for these tables.
¨ To add or drop partitions in a partitioned database and then redistribute table data across the updated partition
group, or to reconfigure a database partition group, you must use a special procedure. Otherwise,
PowerExchange might not be able to resume change data capture properly.
¨ If you alter a column datatype to or from FOR BIT DATA, PowerExchange does not detect the datatype
change. PowerExchange continues to use the datatype that is specified in the existing capture registration.
¨ In a partitioned database, if an UPDATE to a table row changes the partition key and that change causes the
row to move to another partition, PowerExchange processes the UPDATE as two operations: a DELETE and
an INSERT. However, based on the DB2 log information, PowerExchange cannot predictably determine the
order in which to perform the DELETE and INSERT operations. If the INSERT is processed first, both the
original row and the updated row appear on the target until the DELETE is processed.
¨ The maximum length of a row from which PowerExchange can capture change data is 32 KB.
RELATED TOPICS:
¨ “Reconfiguring a Partitioned Database or Database Partition Group” on page 67
Configuring DB2 for CDC

To configure DB2 for Linux, UNIX, or Windows for PowerExchange CDC, perform the following tasks:
1. In the DB2 Control Center Configure Database Logging Wizard, enable archive logging for the DB2 database.
For more information, see the IBM DB2 documentation.
If archive logging is not enabled, PowerExchange issues the error messages PWX-20204 and PWX-20229
during CDC.
2. Set the following user environment variables in any process that runs PowerExchange CDC or the
DTLUCUDB program:
¨ Set DB2NOEXITLIST to ON.
¨ Set DB2CODEPAGE to 1208.
3. Verify that the DB2 source tables are defined with the DATA CAPTURE CHANGES clause.
4. If a table that is selected for change data capture includes columns with a LONG datatype, use the INCLUDE
LONGVAR COLUMNS clause to alter the table so that PowerExchange can capture data for the LONG
columns. Otherwise, PowerExchange might issue the error message PWX-20094 during CDC processing.
58 Chapter 4: DB2 for Linux, UNIX, and Windows Change Data Capture
Configuring PowerExchange for DB2 CDC
The tasks that you perform to configure PowerExchange for DB2 for Linux, UNIX, and Windows CDC depend on
whether you want to use the PowerExchange Logger for Linux, UNIX, and Windows and the extraction mode you
plan to use.
RELATED TOPICS:
Configuring PowerExchange CDC without the PowerExchange Logger

If you plan to run extractions in real-time extraction mode and not use the PowerExchange Logger for Linux, UNIX,
and Windows, complete the following tasks to configure PowerExchange CDC:
1. Create the PowerExchange capture catalog table.

2. Run the DTLUCUDB SNAPSHOT command to initialize the capture catalog table.
3. When you configure the dbmover.cfg file, include the following statements:
¨ CAPT_PATH
¨ CAPT_XTRA
¨ UDB CAPI_CONNECTION
4. In the PowerExchange Navigator, create a capture registration for each source table. The PowerExchange
Navigator generates a corresponding extraction map. Optionally, create a data map if you want to perform
field-level processing.
Tip: Set the Condense option to Part even though you do not plan to use the PowerExchange Logger,
unless you have a specific reason not to do so. This practice prevents having to edit the capture registrations
later if you decide to use the PowerExchange Logger. You might want to set the Condense option to None if
you plan to run both real-time and continuous extractions against tables defined by the same capture
registrations and you do not want the PowerExchange Logger to capture change data for some registered
tables.
If capture registrations already exist for the source tables, delete the existing registrations and extraction
maps and create new ones.
5. Activate the capture registrations. Usually, you do this task after materializing the targets.
Next Step: Configure and start extractions. You must use real-time extraction mode.
RELATED TOPICS:
¨ “Initializing the Capture Catalog Table” on page 61
¨ “Customizing dbmover.cfg for DB2 CDC” on page 61
¨ “Creating the Capture Catalog Table” on page 60

Configuring PowerExchange for DB2 CDC 59

Configuring PowerExchange CDC with the PowerExchange Logger
If you plan to use the PowerExchange Logger for Linux, UNIX, and Windows and run extractions in batch or
continuous extraction mode, complete the following tasks to configure PowerExchange CDC:
1. Create the PowerExchange capture catalog table.

2. Run the DTLUCUDB SNAPSHOT command to initialize the capture catalog table.
3. When you configure the dbmover.cfg file, include the following statements:
¨ CAPT_PATH
¨ CAPT_XTRA
¨ UDB CAPI_CONNECTION
¨ CAPX CAPI_CONNECTION (for continuous extraction mode only)
4. Configure the pwxccl.cfg file for the PowerExchange Logger.

5. In the PowerExchange Navigator, create a capture registration for each DB2 source table. You must select
Part in the Condense drop-down list. The PowerExchange Navigator generates a corresponding extraction
map.
If capture registrations already exist for these tables, delete the existing registrations and extraction maps and
create new ones.
7. Start the PowerExchange Logger.
Next Step: Configure and start extractions. You can use either batch extraction mode or continuous extraction
mode.
RELATED TOPICS:
¨ “Configuring the PowerExchange Logger” on page 27
¨ “Starting the PowerExchange Logger” on page 47
¨ “Creating the Capture Catalog Table” on page 60
¨ “Customizing dbmover.cfg for DB2 CDC” on page 61

Creating the Capture Catalog Table

The PowerExchange capture catalog table stores information about the CDC source tables, column definitions,
and valid DB2 log positions. You must create this table in the same database that contains the source tables from
which change data is captured.
If the database has multiple partitions, the capture catalog table stores positioning information for each partition. If
the database has only a single partition, the capture catalog table still contains positioning information for the
partition.
Use the following DDL to create the capture catalog table:

CREATE TABLE DTLCCATALOG (
VTSTIME TIMESTAMP NOT NULL,
VTSACC INTEGER NOT NULL,
NODENUM SMALLINT NOT NULL,
SEQ INTEGER NOT NULL,
TBSCHEMA VARCHAR(128),
TBNAME VARCHAR(128),
OP VARCHAR(1024) NOT NULL,
PRIMARY KEY(VTSTIME, VTSACC, NODENUM, SEQ) )
;
In this DDL, the table name is DTLCCATALOG. If necessary, you can specify another table name.
Tip: Informatica recommends that you place the PowerExchange capture catalog table in the DB2 catalog
partition.
Initializing the Capture Catalog Table

To initialize the PowerExchange capture catalog table, run the DTLUCUDB utility with the SNAPSHOT command.
You should need to do this task only once.
To specify the command, use the following syntax:

DTLUCUDB SNAPSHOT [DB=database_name] [CCATALOG=capture_catalog_name] [UID=user_id]
[EPWD=encrypted_password] [REPLACE=Y|N]
If the capture catalog table contains existing rows of data, you must set the REPLACE parameter to Y to enable
PowerExchange to overwrite the data. For a new capture catalog table, accept the default of N.
After the snapshot successfully completes, back up the capture catalog table to create a point of consistency for
recovery.
Note: If you run the DTLUCUDB SNAPSHOT command while the DB2 catalog is being updated, the snapshot
fails. If this failure occurs, run the SNAPSHOT command again after the DB2 catalog updates are complete.
Customizing dbmover.cfg for DB2 CDC

In the dbmover.cfg configuration file, include the CAPI connection statement that is specific to DB2 for Linux,
UNIX, and Windows. Also add the other statements that are required for CDC and any optional statements that
you want to use.
The following statements are required for DB2 CDC:
¨ CAPT_PATH. Path to the local directory where the following files reside: CCT file for capture registrations,
CDEP file for application names used in ODBC extractions, and CDCT file for information about
PowerExchange Logger for Linux, UNIX, and Windows log files.
¨ CAPT_XTRA. Path to the local directory for extraction maps.
¨ UDB CAPI_CONNECTION. A named set of parameters that the CAPI uses to connect to the change stream
and control extraction processing for DB2 for Linux, UNIX, and Windows sources.
Add this statement to the dbmover.cfg file on the system where DB2 capture registrations are stored. This
location corresponds to the Location node that you specify when defining a registration group. Usually, this
location is where the source database resides.
If you plan to use continuous extraction mode, you must also define the CAPX CAPI_CONNECTION statement.
To find PowerExchange messages more easily, include the LOGPATH statement. This statement defines a
specific directory for the PowerExchange message log files.
RELATED TOPICS:
¨ “DB2 for Linux, UNIX, and Windows CAPI_CONNECTION Parameters” on page 62

Example Statements
The following statements are typical of those included in a dbmover.cfg for DB2 for Linux, UNIX, and Windows
CDC:
CAPT_PATH=c:/pwxcapt/Vnnn
CAPT_XTRA=c:/pwxcapt/Vnnn/extrmaps
CAPI_CONN_NAME=UDBCC
CAPI_CONNECTION=(NAME=UDBCC
,DLLTRACE=bbbb
,TYPE=(UDB
,CCATALOG=mylib.captcat_tbl
,USERID=db2admin
,PASSWORD=db2admin))
DB2 for Linux, UNIX, and Windows CAPI_CONNECTION Parameters

The UDB CAPI_CONNECTION statement specifies the Consumer API (CAPI) parameters needed for DB2 for
Linux, UNIX, and Windows CDC sources.
Data Sources: DB2 for Linux, UNIX, and

Windows
Required: Yes for DB2 for Linux,
UNIX, and Windows CDC
Syntax:
CAPI_CONNECTION=(
NAME=name,
[TRACE=trace,]
TYPE=(UDB,
[CCATALOG=capture_catalog,]
[DBCONN=database_name,]
[EPWD=encryted_password,]
[MEMCACHE=cache_size,]
[PASSWORD=password,]
[RSTRADV=seconds,]
[SPACEPRI=primary_space,]
[UDBSCHEMA=schema,]
[UPDINT=seconds,]
[UPDREC=num_records,]
[USERID=user_id]
)
)
Parameters:
DLLTRACE=trace_id
NAME=name
TRACE=trace
TYPE=(UDB, ... )
Required. Type of CAPI_CONNECTION statement. For DB2 for Linux, UNIX, and Windows sources, this
value must be UDB.
CCATALOG=capture_catalog
Optional. Name of the PowerExchange capture catalog table in the format creator.table_name.
Default is creator.DTLCCATALOG, where creator is the user ID that is used to connect to the database.
DBCONN=database_name
Optional. A database name that specifies an override database to which to connect for data extraction.
The override database must contain tables and columns that are identical to those in the original
database. The original database name is included in the registration tag names and extraction map
names.
Use this parameter if you want extract change data from another database that is identical to the one
specified in the registration group.
EPWD=encryted_password
Optional. Encrypted password that is used with the database user ID specified in the USERID parameter.
You can create encrypted passwords by using the PowerExchange Navigator.
If you specify the USERID parameter, you must specify either the PASSWORD or EPWD parameter. Do
not specify both PASSWORD and EPWD.
MEMCACHE=cache_size
Optional. Memory cache size, in kilobytes, that PowerExchange allocates to reconstruct complete UOWs.
For each extraction session, PowerExchange keeps all changes for each UOW in the memory cache until
it processes the end-UOW record. If the memory cache is too small to hold all of the changes in a UOW,
PowerExchange spills the changes to a sequential files on disk, called UOW spill files.
Each UOW spill file contains one UOW. A UOW might require multiple UOW spill files to hold all of the
changes for that UOW. If the change stream contains multiple large UOWs and the memory cache is
insufficient, PowerExchange might create numerous UOW spill files.
PowerExchange processes the change stream more efficiently if it does not need to use UOW spill files.
In addition to degrading extraction performance, large numbers of UOW spill files can cause a disk space
shortage.
Important: If the change stream contains only small UOWs, the default value might be sufficient.
However, the default value is often too small to eliminate UOW spill files. Informatica recommends that
so you specify a larger value.

The location in which PowerExchange allocates the UOW spill files varies by operating system, as
follows:
¨ For Linux and UNIX, PowerExchange uses the current directory by default for UOW spill files. To use
a different directory, specify the TMPDIR environment variable.
PowerExchange creates the UOW spill file names by using the operating system tempnam function
with a prefix of dtlq.
Note: The UOW spill files are temporary files that are deleted when PowerExchange closes them.
They are not visible in the directory while open.
¨ For Windows, PowerExchange uses the current directory by default for UOW spill files. To use a
different directory, specify the TMP environment variable.
PowerExchange creates the UOW spill file names by using the Windows _tempnam function with a
prefix of dtlq.
Warning: Because PowerExchange allocates the cache size for each extraction operation, use caution
when coding large values for MEMCACHE. Otherwise, many concurrent extraction sessions might cause
memory constraints.
Default is 1024, or 1 MB.
PASSWORD=password
Optional. Clear text password that is used with the database user ID specified in the USERID parameter.
If you specify the USERID parameter, you must specify either the PASSWORD or EPWD parameter. Do
not specify both PASSWORD and EPWD.
RSTRADV=nnnnn
restart point.
each UOW processed.
SPACEPRI=primary_space
Optional. PowerExchange allocates UOW spill files as temporary files.
Default is 2147483647, or 2 GB.
UDBSCHEMA=schema
Optional. Schema name that overrides the schema name in capture registrations.
UPDINT=seconds
Optional. Minimum number of seconds that PowerExchange must wait after encountering a virtual
timestamp (VTS) in the DB2 log records for a partition before writing a positioning entry to the
PowerExchange capture catalog table. The positioning entry, which is composed of a log sequence
number (LSN) and VTS, indicates a location in the DB2 logs.
Note: The UPDREC minimum number of records must also be met before positioning entries can be
written to the capture catalog table.
Default is 600.
UPDREC=number_records
Optional. Minimum number of DB2 log records that PowerExchange must read for a partition before it
can write a positioning entry to the PowerExchange capture catalog table. The positioning entry, which is
composed of a LSN and VTS, indicates a location in the DB2 logs.
Note: The UPDINT minimum wait period must also be met before positioning entries can be written to the
capture catalog table.
Default is 10000.
USERID=user_id
Optional. Database user ID. The user ID must have SYSADM or DBADM authority.
If you specify this parameter, you must also specify either the PASSWORD or EPWD parameter.
Using a DB2 Data Map

If you want PowerExchange to perform field-level processing on some records in a DB2 for Linux, UNIX, and
Windows source table, you must use a data map.
For example, in some DB2 environments, a table can contain a single column that stores an array of fields in a
format that is not consistent with the column datatype, such as a CHAR or VARCHAR column that stores multiple
packed data fields. You can use an expression to modify this data before PowerCenter replicates it to a target.
Also, if you add a user-defined field to a table in record view, you can build an expression to populate it. In the
PowerExchange Navigator, you can define expressions only for data maps.
You might have data maps available for your source tables if you used PowerExchange bulk data movement to
materialize your data targets. Bulk data movement requires data maps. You can use the bulk data maps for CDC if
you merge them with the extraction maps for your data sources. The PowerExchange Navigator automatically
generates an extraction map when you create a capture registration. Alternatively, you can manually add an
extraction map.
Note: The field names in the data map must match the actual column names, as indicated in the DB2 capture
registration.
Using a DB2 Data Map 65

Task Flow for DB2 Data Map Use
Perform the following tasks to use a DB2 data map for change data capture:
1. In the PowerExchange Navigator, create a capture registration for the DB2 source table.
2. Create a DB2 data map for the same DB2 source table if one is not available from a previous bulk data
movement operation.
3. Merge the DB2 data map with the extraction map for the table.
4. Perform a row test on the merged extraction map.
RELATED TOPICS:
¨ “Testing a Change Data Extraction” on page 126
Managing DB2 CDC

You might need to stop DB2 for Linux, UNIX, and Windows CDC for source tables occasionally, for example, to
change the table definitions.
Stopping DB2 CDC

You might need to stop change data capture for a DB2 source table to perform troubleshooting or routine
maintenance tasks, such as maintenance on the capture catalog table or redistribution of table data across
reconfigured database partitions.
To stop change data capture, use one of the following methods:
¨ Open the capture registration for a source table, and change the Status value from Active to History.
Warning: After you set the status of a capture registration to History, you cannot activate the registration
again. This status change permanently stops change data capture based on the capture registration.
¨ To temporarily stop change data capture, alter the DB2 table to specify the DATA CAPTURE NONE clause:
ALTER owner.table_name DATA CAPTURE NONE
When DATA CAPTURE NONE is specified, DB2 no longer writes changes to the DB2 log files in expanded
format. Because CDC requires expanded format, PowerExchange can no longer capture change data for the
table from the log files. If you set it back to DATA CAPTURE CHANGES, you might need to rematerialize the
targets.
RELATED TOPICS:
¨ “Stopping PowerCenter CDC Sessions” on page 142
Changing a DB2 Source Table Definition

Occasionally, you might need to change the definition of a DB2 for Linux, UNIX, and Windows source table that is
registered for change data capture. If your metadata changes affect the columns from which change data is
captured, use this procedure to enable PowerExchange to switch to the updated table definition, while preserving
access to previously captured data.
Perform this procedure whenever you add, alter, or drop columns for which change data is captured. You do not
need to perform this procedure if you are selectively capturing change data for a subset of columns and none of
the selected columns are affected by the metadata changes.
Tip: If you no longer need to capture change data from a column in a table, you can remove the column from the
extraction map without changing the capture registration. Change data for that column is still captured but is not
extracted.
To change a DB2 source table definition:
1. Stop DELETE, INSERT, and UPDATE activity against the table.

2. Verify that any change data that was captured under the previous table definition has completed extraction
processing. Then stop all workflows that extract change data for the table.
3. In the PowerExchange Navigator, open the original capture registration and set its status to History.
Alternatively, if you need to add or delete existing columns, right-click the capture registration and click
Amend Columns. You can then add or delete columns, as needed. This action creates a new version of the
capture registration that has a status of Inactive.
Note: PowerExchange does not capture change data based on capture registrations that have a status of
History or Inactive.
4. Use DDL to make the table changes.
5. In the PowerExchange Navigator, create a new capture registration that reflects the metadata changes and
set its status to Active.
Alternatively, if you created a new version of the original capture registration by amending columns, you can
add any new columns that you defined. Then set the capture registration status to Active. Also, edit the
associated extraction map to point to the new capture registration version. Right-click the associated
extraction map and click Amend Capture Registrations.
PowerExchange uses the newly activated capture registration for change data capture.
6. If necessary, change the target table definition to reflect the source table metadata changes.
7. In PowerCenter Designer, import the altered source and target tables. Edit the mapping if necessary.
8. If necessary, rematerialize the target tables. After materialization completes, create new restart tokens.
9. Re-enable DELETE, INSERT, and UPDATE activity against the table.
10. Restart extraction processing.
RELATED TOPICS:
¨ “Creating Restart Tokens for Extractions” on page 135
Reconfiguring a Partitioned Database or Database Partition Group

In a DB2 for Linux, UNIX, and Windows partitioned database environment, you might need to perform the following
reconfiguration tasks:
¨ Add a new partition to a partitioned database, or drop an existing partition. Then reconfigure the database
partition group or groups to reflect the change.
¨ Reconfigure a database partition group by adding or removing existing partitions.
Typically, after making these types of changes, you run the DB2 REDISTRIBUTE DATABASE PARTITION
GROUP command to redistribute table data among the partitions in the updated database partition group.
If PowerExchange change data capture is active in the partitioned database environment, you must use the
following procedure to properly resume change data capture after making the reconfiguration changes.
Managing DB2 CDC 67

Adding or Dropping Database Partitions
Use the following procedure to create a new partition in a partitioned database or to drop an existing partition, and
then update the appropriate database partition group for the change:
1. In PowerCenter, stop all CDC sessions that extract change data for the tables in the partitioned database
instance.
2. For each table for which the DATA CAPTURE CHANGES clause is specified, specify DATA CAPTURE
NONE.
Note: This step temporarily disables DB2 capture of changes to its log files. If you do not perform this step,
DB2 records the data redistribution changes that result from the RESTRIBUTE command as regular change
data activity.
3. Execute the SQL for adding the new database partition or for dropping an existing partition.
4. Execute the ALTER DATABASE PARTITION GROUP SQL to add the new partition to or remove the dropped
partition from the appropriate database partition group.
5. Run the DB2 REDISTRIBUTE DATABASE PARTITION GROUP command to redistribute table data among
the partitions in the altered database partition group.
6. Back up the PowerExchange capture catalog table.
7. Run the PowerExchange DTLUCUDB SNAPUPDT command. Set the REPLACE option set to Y. This step
updates the PowerExchange capture catalog table to reflect the reconfigured partitioned database.
Tip: Informatica recommends that you first perform a test run with the REPLACE option set to N.
8. For each table for which you specified DATA CAPTURE NONE in step 2, reinstate the DATA CAPTURE
CHANGES clause.
9. Restart the PowerCenter CDC sessions to resume extraction processing.
RELATED TOPICS:
Reconfiguring a Database Partition Group

Use the following procedure to add a partition to or remove a partition from a database partition group without
changing the partitioning of the partitioned database instance:
1. In PowerCenter, stop all CDC sessions that extract change data for the tables in the partitioned database
instance.
2. For each table for which the DATA CAPTURE CHANGES clause is specified, specify DATA CAPTURE
NONE.
Note: This step temporarily disables DB2 capture of changes to its log files. If you do not perform this step,
DB2 records the data redistribution changes that result from the RESTRIBUTE command as regular change
data activity.
3. Execute the ALTER DATABASE PARTITION GROUP SQL to add the new partition to or remove the dropped
partition from the appropriate database partition group.
4. Run the DB2 REDISTRIBUTE DATABASE PARTITION GROUP command to redistribute table data among
the partitions in the altered database partition group.
5. For each table for which you specified DATA CAPTURE NONE in step 2, reinstate the DATA CAPTURE
CHANGES clause.
6. Restart the PowerCenter CDC sessions to resume extraction processing.
DB2 for Linux, UNIX, and Windows CDC
Troubleshooting
If you encounter the following issue when running DB2 for Linux, UNIX, and Windows CDC, attempt the solution
that is described. If you cannot resolve the problem, contact Informatica Global Customer Support.
Workaround for SQL1224 Error on AIX

On AIX systems only, you might receive the following PowerExchange message for a DB2 SQL1224 error when
you connect locally to a DB2 database that has multiple other local connections:
PWX-20604 State=08001, Code=-1224, Msg=[IBM][CLI Driver] SQL1224N A database agent could not be
started to service a request, or was terminated as a result of a database system shutdown or a force
command. SQLSTATE=55032.
To circumvent this problem, implement a loopback TCP/IP connection for the local DB2 database. The database
can then function as a remote client that uses TCP/IP instead of interprocess communications (IPC) over shared
memory.
To implement a loopback connection without changing the database alias that users enter for database
connection, issue the following DB2 commands:
db2 catalog tcpip node node_name1 remote server_name1 server port_number1
db2 uncatalog database database_name1
db2 catalog database database_name1 at node node_name1
db2 catalog database database_name1 as database_alias1
db2 catalog database database_alias1 as database_name1 at node node_name1
For more information about these commands, see your IBM DB2 documentation.
IBM APARs for Specific Issues

If you encounter the issues documented in the following IBM APARs, go to the IBM Web site for more information
or apply the appropriate FixPak for your DB2 version.
DB2 for Linux, UNIX, and Windows 9.5:
¨ The following issue can cause invalid PowerExchange capture registrations, which include character columns
with an incorrect code page:
JR30422: "ALTER TABLE ALTER COLUMN" STATEMENT DOES NOT ALTER THE CODEPAGE COLUMN IN THE
SYSCAT.COLUMNS VIEW.
To resolve this issue, search the IBM Web site for the latest information about this APAR.
DB2 for Linux, UNIX, and Windows 9.1:
¨ The following issue can result in a SQL error message:

IY87631: PESSIMISTIC LOCKING FOR CLI SQL_CONCUR_LOCK NO LONGER WORKING IN V8
The SQL error message is:
SQL0644N Invalid value specified for keyword "CONCURRENCY" in statement "ATTRIBUTE-STRING".
SQLSTATE=42615
To resolve this issue, apply DB2 9.1 FixPak 1 or later.
¨ The following issue can cause invalid PowerExchange capture registrations that include character columns with
an incorrect code page:
JR30420: "ALTER TABLE ALTER COLUMN" STATEMENT DOES NOT ALTER THE CODEPAGE COLUMN IN THE
SYSCAT.COLUMNS VIEW.
To resolve this issue, search the IBM Web site for the latest information about this APAR.
DB2 for Linux, UNIX, and Windows CDC Troubleshooting 69

CHAPTER 5
Microsoft SQL Server Change Data

Capture
¨ Microsoft SQL Server CDC Overview, 70
¨ Planning for SQL Server CDC, 71
¨ Configuring SQL Server for CDC, 73
¨ Configuring PowerExchange for SQL Server CDC, 74
¨ Managing SQL Server CDC, 78
Microsoft SQL Server CDC Overview

PowerExchange uses SQL Server transactional replication to capture change data from SQL Server distribution
databases. PowerExchange uses the PowerExchange Client for PowerCenter (PWXPC) to coordinate with
PowerCenter to move the captured change data to one or more targets.
For CDC to work, you must enable SQL Server Replication on the system from which change data is to be
captured. If your database has a high volume of change activity, you should use a distributed server as the host of
the distribution database.
To configure CDC in PowerExchange, you must define a capture registration for each source table. In the capture
registration, you can select a subset of columns for which to capture data. PowerExchange generates a
corresponding extraction map.
If you want to use the PowerExchange Logger for Linux, UNIX, and Windows to capture change data and write it
to PowerExchange Logger log files, configure the PowerExchange Logger. The change data is then extracted from
the PowerExchange Logger log files. Benefits of the PowerExchange Logger include fewer database accesses
and faster CDC restart.
PowerExchange works with PowerCenter to extract change data from the SQL Server distribution database or
PowerExchange Logger log files and load that data to one or more targets.
RELATED TOPICS:
70
Planning for SQL Server CDC
Before you configure SQL Server change data capture (CDC), verify that the following prerequisites and user
authority requirements are met. Also, review the restrictions so that you can properly configure CDC.
SQL Server CDC Prerequisites

PowerExchange CDC has some SQL Server prerequisites.
These prerequisites are:
¨ PowerExchange CDC requires an edition of Microsoft SQL Server 2000 or later that supports transactional
replication. You must configure and enable transactional replication on the source system to participate in CDC.
¨ If you use Microsoft SQL Server 2008, install the Microsoft SQL Server 2005 Backward Compatibility
components if have not done so. You can download these components from the Microsoft Web site.
¨ The Microsoft SQL Server Agent and Log Reader Agent must be running on the Windows machine from which
change data is extracted. Usually, the SQL Server Agent remains running after it is initially started. For more
information, see your SQL Server documentation.
¨ Each source table in the distribution database must have a primary key.
¨ If the PowerExchange Navigator does not reside on the same machine as the Microsoft SQL Server software,
you must install the SQL Server client components on the PowerExchange Navigator machine.
Required User Authority for SQL Server CDC

PowerExchange CDC requires the following user authority levels:
¨ To create capture registrations in the PowerExchange Navigator, you must be a member of the SQL Server
sysadmin server role.
¨ To run change data extractions against a SQL Server distribution database, you must have read access to that
database.
If you do not specify a user ID and password, the PowerExchange Navigator and your extraction processes
attempt to use your Windows user ID and password to connect to the SQL Server distribution database.
Datatypes Supported for SQL Server CDC

This topic identifies the SQL Server datatypes that PowerExchange supports for CDC.
The following table lists the datatypes and indicates whether they are supported for CDC:
Datatype Supported for CDC? Comments
bigint Yes
binary Yes
bit Yes
char Yes
date No This datatype was introduced in SQL Server 2008.
Planning for SQL Server CDC 71

datetime Yes
datetime2 No This datatype was introduced in SQL Server 2008.
datetimeoffset No This datatype was introduced in SQL Server 2008.
decimal Yes
float Yes
geography No This datatype was introduced in SQL Server 2008.
geometry No This datatype was introduced in SQL Server 2008.
hierarchyid No This datatype was introduced in SQL Server 2008.
image1 No Use varbinary(MAX) instead.
int Yes
money Yes
nchar Yes
ntext1 No Use nvarchar(MAX) instead.
numeric Yes
nvarchar Yes
real Yes
smalldatetime Yes
smallint Yes
smallmoney Yes
sql_variant No PowerExchange does not capture change data for

sql_variant columns but does capture change data
for other columns in the same table.
text1 No Use varchar(MAX) instead.
time No This datatype was introduced in SQL Server 2008.
timestamp Yes
tinyint Yes
uniqueidentifier Yes PowerCenter imports the uniqueidentifier datatype

as a varchar datatype of 38 characters.
72 Chapter 5: Microsoft SQL Server Change Data Capture

user-defined datatypes (UDTs) Yes PowerExchange treats a UDT in the same way as
the datatype on which the UDT is based.
varbinary Yes
varchar Yes
xml Yes PowerExchange treats this datatype as

varchar(MAX).
1. PowerExchange might not be able to capture change data for columns that have the datatypes of image, ntext, or text
because of SQL Server transactional replication restrictions on these types of columns. Instead, use the alternative datatypes
that Microsoft recommends, as shown in the Comments column.
SQL Server CDC Restrictions

The following restrictions apply to SQL Server CDC:
¨ PowerExchange does not capture change data for SQL Server system tables.
¨ The maximum length of a row for which PowerExchange can capture and process change data is 32 KB.
¨ PowerExchange does not capture the user ID that is associated with the original transaction that updated the
database.
¨ The timestamp that PowerExchange records for each captured change indicates when the change was
captured, not when the original transaction occurred.
¨ PowerExchange does not capture change data for derived columns that are not persisted. SQL Server
computes values for these columns at run-time based on an expression but does not store the values in a table.
¨ SQL Server publishes deferred updates to SQL Server tables as DELETEs followed by INSERTs rather than as
UPDATEs. Consequently, PowerExchange propagates deferred updates as DELETEs followed by INSERTs,
even if you select AI for the Image Type attribute in the CDC connection. PowerExchange does not include
before image (BI) and change indicator (CI) information in DELETE and INSERT operations. For more
information about deferred updates, see your Microsoft SQL Server documentation.
Configuring SQL Server for CDC

You must perform a few configuration tasks to prepare SQL Server for PowerExchange change data capture
(CDC).
If your SQL Server tables have a high level of update activity, use a distributed server as the host of the
distribution database from which change data is captured. This practice prevents competition between
PowerExchange CDC and your production database for CPU use and disk storage.
Configuring SQL Server for CDC 73

To configure SQL Server for PowerExchange CDC, perform the following tasks:
1. Start the SQL Server Agent and Log Reader Agent if they are not running. For more information, see your
Microsoft SQL Server documentation.
2. Configure and enable SQL Server transactional replication. For more information, see your Microsoft SQL
Server documentation.
Tip: The default transactional retention period at the Distributor is 72 hours. If you are use the
PowerExchange Logger, accept this default retention period. If you do not use the PowerExchange Logger,
Informatica recommends that you increase the retention period to 14 days. However, you might need to a
lower value if you have a high volume of transactions or space constraints.
3. Verify that each source table in the distribution database has a primary key.
Configuring PowerExchange for SQL Server CDC

The tasks that you perform to configure PowerExchange for change data capture (CDC) depend on whether you
want to use the PowerExchange Logger for Linux, UNIX, and Windows and the extraction mode you plan to use.
RELATED TOPICS:
Configuring PowerExchange CDC without the PowerExchange Logger

and Windows, complete the following tasks to configure PowerExchange CDC:
1. When you configure the dbmover.cfg file, define the following statements:
¨ CAPT_PATH
¨ CAPT_XTRA
¨ MSQL CAPI_CONNECTION
2. In the PowerExchange Navigator, create a capture registration for each SQL Server source table. The
PowerExchange Navigator generates a corresponding extraction map.
unless you have a particular reason not to do so. This practice prevents having to change the capture
registrations later if you decide to use the PowerExchange Logger. You might want to set the Condense
option to None if you run both real-time and continuous extractions against tables defined by the same
capture registrations and do not want the PowerExchange Logger to capture change data for certain
registered tables.
create new ones.
The PowerExchange Navigator generates a corresponding extraction map for each capture registration.
RELATED TOPICS:
¨ “Customizing dbmover.cfg for SQL Server CDC” on page 75

Configuring PowerExchange CDC with the PowerExchange Logger

If you plan to run extractions in batch or continuous extraction mode and use the PowerExchange Logger for
Linux, UNIX, and Windows, complete the following tasks to configure PowerExchange CDC:
1. When you configure the dbmover.cfg file, define the following statements:
¨ CAPT_PATH
¨ CAPT_XTRA
¨ MSQL CAPI_CONNECTION

3. In the PowerExchange Navigator create a capture registration for each SQL Server source table. You must
set the Condense option to Part. The PowerExchange Navigator generates a corresponding extraction map.
create new ones.
mode.
RELATED TOPICS:
¨ “Customizing dbmover.cfg for SQL Server CDC” on page 75
Customizing dbmover.cfg for SQL Server CDC

In the dbmover.cfg configuration file, include the CAPI connection statement that is specific to SQL Server. Also
add the other statements that are required for CDC and any optional statements that you want to use.
The following statements are required for SQL Server CDC:
¨ CAPT_PATH. Path to the local directory that stores the following files for CDC: CCT file for capture
registrations, CDEP file for application names used in ODBC extractions, and CDCT file for information about
PowerExchange Logger for Linux, UNIX, and Windows log files.
¨ CAPT_XTRA. Path to the local directory that stores extraction maps.
¨ MSQL CAPI_CONNECTION. A named set of parameters that the CAPI uses to connect to the change stream
and control extraction processing for SQL Server CDC. Add this statement to the dbmover.cfg file on the
system where SQL Server capture registrations are stored. This location corresponds to the Location node
that you specify when defining a registration group. Usually, this location is where the source database resides.
If you plan to use the PowerExchange Logger and continuous extraction mode, you must also define the CAPX
CAPI_CONNECTION statement.
Configuring PowerExchange for SQL Server CDC 75

To find PowerExchange messages more easily, include the LOGPATH statement. This statement defines a
specific directory for the PowerExchange message log files.
RELATED TOPICS:
¨ “Microsoft SQL Server CAPI_CONNECTION Parameters” on page 76
Example Statements
The following statements are typical of those included in a dmover.cfg for SQL Server CDC:
LOGPATH="C:\Informatica\PowerExchangeVnnn\Logs"
CAPT_XTRA="C:\Informatica\PowerExchangeVnnn\Capture\camaps"
CAPT_PATH="C:\Informatica\PowerExchangeVnnn\Capture"
CAPI_CONN_NAME=CAPIMSSC
CAPI_CONNECTION=(NAME=CAPIMSSC
,TYPE=(MSQL,DISTSRV=AUX159908\PWXPC
,DISTDB=distribution
,RSTRADV=30))
Note: You must use non-curly double quotation marks around values that include a space.
Microsoft SQL Server CAPI_CONNECTION Parameters

The MSQL CAPI_CONNECTION statement specifies the Consumer API (CAPI) parameters needed for Microsoft
SQL Server CDC sources.
Data Sources: Microsoft SQL Server

Required: Yes for Microsoft SQL
Server CDC
Syntax:
CAPI_CONNECTION=(
NAME=name,
[TRACE=trace,]
TYPE=(MSQL,
DISTDB=distribution_database,
DISTSRV=distribution_server,
[DWFLAGS=flag1flag2flag3,]
[EOF={N|Y},]
[POLWAIT=seconds,]
[RSTRADV=seconds]
)
)
Parameters:
DLLTRACE=trace_id
NAME=name

TRACE=trace
TYPE=(MSQL, ... )
Required. Type of CAPI_CONNECTION statement. For Microsoft SQL Server sources, this value must be
MSQL.
DISTDB=distribution_database
Required. Name of the distribution database.
DISTSRV=distribution_server
Required. Network name of the server that hosts the distribution database.
Important: This name is different from the network name of the instance if the distribution database
resides on a different server.
DWFLAGS=flag1flag2flag3
Optional. Series of three positional parameters that control whether processing stops or continues when
data loss, truncation, or schema changes occur.
Enter the following positional parameters:
¨ flag1. Controls whether PowerExchange stops a change data extraction when data of an unexpected
length is retrieved from the distribution database. Enter Y to continue processing or N to stop
processing.
¨ flag2. Controls whether PowerExchange stops a change data extraction when a schema change is
detected. Enter Y to continue processing or N to stop processing.
¨ flag3. Controls whether PowerExchange stops a change data extraction when the requested start
sequence is not found in the transaction log. Enter Y to continue processing or N to stop processing.
Default is NNN.
EOF={N|Y}
Optional. Controls whether PowerExchange stops change data extractions when the end-of-log (EOL) is
reached.
Enter one of the following options:
¨ N. PowerExchange does not stop change data extractions when EOL is reached.
¨ Y. PowerExchange stops change data extractions when EOL is reached.
Because this parameter affects all users of the AS4J CAPI_CONNECTION statement, Informatica
recommends that you use one of the following alternative methods to stop change data extractions at
EOL:
¨ For CDC sessions that use real-time extraction mode, enter 0 for the Idle Time attribute of the PWX
MSSQL CDC Real Time application connection.
¨ For PowerExchange Logger for Linux, UNIX, and Windows, enter 1 for the COLL_END_LOG
statement in the pwxccl.cfg configuration file.
¨ For CDC sessions that use ODBC connections, enter 0 for the WAITTIME parameter in the ODBC
data source.
Default is N.
Configuring PowerExchange for SQL Server CDC 77

MEMCACHE=cache_size
Optional. Memory cache size, in kilobytes, that PowerExchange allocates to cache a single change.
Default is 248.
POLWAIT=seconds
Optional. Time interval, in seconds, that PowerExchange waits after reaching the end of current data
before polling for new data.
Default is 1.
RSTRADV=nnnnn
restart point.
each UOW processed.
Managing SQL Server CDC

You might need to stop CDC for source tables occasionally, for example, to change the table definitions.
Disabling Publication of Change Data for a SQL Server Source

You can disable publication of change data for a SQL Server source. For example, you might disable publication
to perform some database maintenance, change the table definition, or avoid capturing unwanted changes.
u Open the capture registration for the table, and change the Status setting from Active to History.
This action disables publication of the SQL Server article for the table to the distribution database, which
causes change capture to stop.
Warning: After the registration status is set to History, you cannot activate the registration for CDC use again.

Changing a SQL Server Source Table Definition
If you change the definition of a SQL Server source table that is registered for change data capture, use this
procedure to enable PowerExchange to use the updated table definition and preserve access to previously
captured data. Table definition changes include adding, altering, or dropping columns.
Tip: If you no longer need to capture change data from a column in a table, you can remove the column from the
extraction map without changing the capture registration. Change data for that column is still captured but is not
extracted.
To change a SQL Server source table definition:

3. Delete the capture registration and extraction map.
4. Use DDL to change the table definition in SQL Server.
set its status to Active. PowerExchange creates a corresponding extraction map.
The newly activated capture registration becomes eligible for change data capture.
7. In the PowerCenter Designer, import the altered source and target definitions. Edit the mapping if necessary.
9. Create new restart tokens for the altered table.
11. Cold start the extraction workflows.
Managing SQL Server CDC 79

CHAPTER 6
Oracle Change Data Capture with

Oracle LogMiner
¨ Overview of Oracle LogMiner CDC, 80
¨ Planning for Oracle LogMiner CDC, 81
¨ Oracle Configuration for LogMiner CDC, 83
¨ PowerExchange Configuration for Oracle LogMiner CDC, 88
¨ Management of Oracle LogMiner CDC, 102
Overview of Oracle LogMiner CDC

PowerExchange can use Oracle LogMiner to read change data from Oracle redo logs. To move the change data to
one or more targets, PowerExchange uses the PowerExchange Client for PowerCenter (PWXPC) in conjunction
with PowerCenter.
To implement Oracle LogMiner CDC, you need to perform configuration tasks in Oracle, PowerExchange, and
PowerCenter.
In Oracle, ensure that ARCHIVELOG mode with global minimal supplemental logging is enabled so that change
data can be retrieved from archived redo logs. Also, ensure that a copy of the Oracle online catalog exists in the
archived redo logs. PowerExchange requires a copy of the catalog to determine restart points for change data
extraction processing.
In PowerExchange, define a capture registration for each source table. In the capture registration, you can select a
subset of columns for which to capture data. PowerExchange generates a corresponding extraction map.
If you want to use the PowerExchange Logger for Linux, UNIX, and Windows, also configure the PowerExchange
Logger. The PowerExchange Logger can capture change data from Oracle redo logs and write only the successful
units of work (UOWs), in chronological order based on commit time, to PowerExchange Logger log files. The
change data is then extracted from the PowerExchange Logger log files in either continuous extraction mode or
batch extraction mode. Benefits of using the PowerExchange Logger include fewer database accesses, faster
CDC restart, and no need to prolong retention of the Oracle redo files for change capture.
Note: Informatica strongly recommends that you use the PowerExchange Logger for Oracle LogMiner CDC. If you
use real-time extraction mode without the PowerExchange Logger, PowerExchange starts a separate Oracle
LogMiner session for each extraction session. Running multiple, concurrent sessions can significantly degrade
performance of the system where LogMiner runs.
80
PowerExchange works with PowerCenter to extract change data from Oracle redo logs or PowerExchange Logger
log files and load that data to one or more targets.
RELATED TOPICS:
Planning for Oracle LogMiner CDC

Before you configure Oracle change data capture, review the following restrictions, requirements, and
performance considerations.
Requirements and Restrictions for Oracle LogMiner CDC

The following restrictions and requirements apply to Oracle LogMiner CDC:
¨ The Oracle instance must be running in ARCHIVELOG mode.
¨ Oracle global minimal supplemental logging must be enabled.
¨ A copy of the Oracle catalog must exist in the Oracle archived redo logs.
¨ Oracle LogMiner continuous mining reads archived redo logs only from the directory to which they were
originally written.
¨ If you truncate Oracle source tables from which change data is captured, or if you drop and re-create source
tables, PowerExchange cannot continue to extract change data for these tables. In these situations, you must
rematerialize the corresponding targets.
¨ If PowerExchange CDC is not installed on the same machine as the Oracle instance, configure a TNS entry on
the client machine with SERVER=DEDICATED in the CONNECT_DATA section of the connect descriptor. This
specification is also required if the network is configured for Multi-Threaded Server (MTS) mode.
¨ PowerExchange requires the Oracle Client binaries. When you install Oracle, the Client binaries are installed
by default. To use SQL*Net connectivity on a machine that does not have an installed Oracle instance, you
must install the Oracle Client.
¨ The maximum length of a row for which PowerExchange can capture and process change data is 32 KB.
Datatypes Supported for Oracle LogMiner CDC

PowerExchange uses Oracle LogMiner to retrieve changes from the Oracle redo logs. Oracle does not log, or
does not completely log, data with some datatypes in the Oracle redo logs. Consequently, PowerExchange cannot
retrieve change data for columns that have these datatypes.
The following table identifies the Oracle datatypes that PowerExchange supports for Oracle LogMiner CDC:
BFILE No Data for columns that have this datatype are not
completely logged in the Oracle redo logs and
cannot be captured.
BINARY_DOUBLE Yes
Planning for Oracle LogMiner CDC 81

BINARY_FLOAT Yes
CHAR Yes
DATE Yes
FLOAT Yes
LOBs No
LONG No
LONG RAW No
NCHAR Yes For CDC support of this datatype, you must have
PowerExchange 8.5 or later.
NUMBER Yes PowerExchange handles NUMBER columns as

follows:
- Numbers with a scale of 0 and a precision value
less than 10 are treated as INTEGER.
- Numbers with a defined precision and scale are
treated as NUMCHAR.
- Numbers with an undefined precision and scale
are treated as DOUBLE.
NVARCHAR2 Yes For CDC support of this datatype, you must have
PowerExchange 8.5 or later.
RAW Yes
TIMESTAMP Yes
TIMESTAMP WITH TIME ZONE No
TIMESTAMP WITH LOCAL TIME ZONE No
VARCHAR2 Yes
SQL*Loader Restrictions
PowerExchange CDC can capture data that was loaded into Oracle tables by the SQL*Loader utility. However, the
following restrictions apply:
¨ The load type must be conventional path. PowerExchange cannot capture data that was loaded by a direct
path load because Oracle LogMiner does not support direct path loads.
¨ The load method should be Insert, Append, or Replace. Do not use Truncate. Truncate causes SQL*Loader to
issue TRUNCATE TABLE DDL. Because PowerExchange does not capture DDL, it cannot capture any row
deletions that result from TRUNCATE TABLE DDL.
82 Chapter 6: Oracle Change Data Capture with Oracle LogMiner

Performance Considerations for Oracle LogMiner CDC
The following considerations pertain to PowerExchange CDC performance:
¨ Use real-time extraction mode only if you run very few concurrent change data extractions. PowerExchange
CDC creates an Oracle LogMiner session for each real-time extraction. Because LogMiner sessions are
resource intensive, they can impact overall system performance. Instead, use continuous extraction mode. For
continuous extraction mode, PowerExchange extracts change data from PowerExchange Logger log files.
¨ If you use continuous extraction mode, minimize the size of the CDCT file. The CDCT file contains information
about the PowerExchange Logger log files. PowerExchange reads the CDCT file each time the interval that is
specified in the FILEWAIT parameter of the CAPX CAPI_CONNECTION statement elapses. If a CDCT file is
large, PowerExchange read operations can result in a high level of I/O activity, increased use of system
resources, and increased extraction latency. To manage the CDCT file size, use the COND_CDCT_RET_P
statement in the pwxccl.cfg configuration file for the PowerExchange Logger for Linux, UNIX, and Windows.
Oracle Configuration for LogMiner CDC

PowerExchange provides sample script files to help you configure Oracle for PowerExchange CDC.
Configuration Script Files

To configure Oracle for CDC, use the sample Oracle configuration script files in the PowerExchange installation
directory.
PowerExchange provides the following script files for RAC and non-RAC environments:
oracapt.sql
Configures Oracle for CDC in a non-RAC environment.
oracapt_rac.sql
Configures Oracle for CDC in an RAC environment. PowerExchange supports CDC in RAC environments only
for Oracle 10g Release 2 and later.
Each script file contains sample SQL statements for performing the necessary configuration tasks. Before running
any of the SQL statements, read the comments in the script file. The comments provide important information.
Use the script file that is appropriate for your environment to perform the following configuration tasks:
¨ Grant required Oracle privileges.
¨ Enable ARCHIVELOG mode.
¨ Enable global minimal supplement logging.
¨ Configure Oracle LogMiner.
¨ Copy the Oracle catalog to the archived redo logs.
¨ Set the transaction_auditing parameter to “True,” if you run an Oracle version earlier than 10.2.01.
Configuring Oracle for LogMiner CDC

This section describes steps for configuring Oracle for LogMiner CDC. For sample SQL and DDL, refer to the
oracapt.sql file.
Oracle Configuration for LogMiner CDC 83

Step 1. Specify an Archive Log Destination
Edit your init.ora file to specify the archive log destination and file-name format. For more information, see your
Oracle database administrator's guide.
Alternatively, if you use a server parameter file (spfile), issue the following SQL statements to indicate the archive
log destination:
CONNECT SYS/sys_pwd AS SYSDBA;
ALTER SYSTEM SET
log_archive_dest_1 = 'location=/oracle_path/arch'
SCOPE=SPFILE;
Step 2. Set the Oracle Compatible Parameter (Oracle 9.2.0)

If you use Oracle 9.2.0 and the "compatible" parameter is not specified in the init.ora or spfile file, or if the
"compatible" parameter is set to an Oracle version earlier than 9.2.0, you must set this parameter to 9.2.0.
To set this parameter, issue the following SQL statement:

ALTER SYSTEM SET compatible=?9.2.0? SCOPE=SPFILE;
Step 3. Set the Oracle transaction_auditing Parameter

If you use an Oracle version earlier than 10.1.0.1, verify that the transaction_auditing parameter is set to "True" in
the init.ora or spfile file. This setting is required for Oracle CDC to work properly.
To set this parameter in the spfile file, execute the following SQL statement in an SQL*Plus session:
Alter SYSTEM SET transaction_auditing=TRUE SCOPE=SPFILE;
For more information, see the oracapt.sql or oracapt_rac.sql file.
If you run Oracle 9.2.0.6 or 10.1.0.4, install the appropriate patch for your release instead. You can find the patch
by searching My Oracle Support (formerly MetaLink) Knowledge Base for bug report 3456259.
Step 4. Enable ARCHIVELOG Mode

For CDC, Oracle must be running in ARCHIVELOG mode.
By default, ARCHIVELOG mode is not enabled.
To enable ARCHIVELOG mode, issue the following statements:

SHUTDOWN IMMEDIATE;
STARTUP MOUNT;
ALTER DATABASE ARCHIVELOG;
ALTER DATABASE OPEN;
SHUTDOWN IMMEDIATE;
STARTUP;
Tip: Back up your database after both SHUTDOWN commands.
If you use the Oracle init.ora initialization parameter file, you must edit the appropriate parameters in this file to
identify the archive log destination and file name format. For more information, see the Oracle database
administrator’s guide for your Oracle version.
If you use a server parameter file (spfile), you must execute some ALTER SYSTEM SET SQL. The specific SQL
and configuration steps vary for RAC and non-RAC environments and are described in the oracapt.sql and
oracapt_rac.sql files.
Step 5. Stop and Restart the Oracle Database

If you set the ARCHIVELOG mode or the "compatible" or "transaction_auditing" parameter, you must stop and
restart the Oracle instance for your changes to take effect.

For more information, see the oracapt.sql file.
Step 6. Grant User Privileges Required for Oracle LogMiner CDC

To extract change data from Oracle redo logs, a CDC user must have specific Oracle system and object
privileges. You can either use an existing user who has the required authority as the CDC user, or create a user
and grant the required privileges to that user.
The oracapt.sql and oracapt_rac.sql configuration script files contain the required SQL GRANT statements. Edit
this SQL, as needed, for your environment.
The following table identifies the minimum system privileges that Oracle CDC users must have:
System Privilege Oracle Release Description
ALTER ANY TABLE All Required for users that create capture registrations and allow
PowerExchange to automatically run the DDL that is
generated for creating a supplemental log group at
registration completion.
CONNECT All Required for users that extract Oracle CDC data in real time
and for PowerExchange Logger tasks.
LOCK ANY TABLE All If you specify GENRLOCK=Y in the ORCL

CAPI_CONNECTION statement of the dbmover.cfg file, you
must either grant the LOCK ANY TABLE system privilege or
grant the SELECT object privilege on each table that is
registered for change data capture.
SELECT ANY TRANSACTION 10g and later Required for users who extract Oracle CDC data in real time
and for PowerExchange Logger tasks.
The following table identifies the minimum object privileges that Oracle CDC users must have:
Object Name Object Privilege
Source tables If you specify GENRLOCK=Y in the

ORCL CAPI_CONNECTION
statement of the dbmover.cfg file, you
must either grant the LOCK ANY
TABLE system privilege or grant the
SELECT object privilege on each
table that is registered for change
data capture.
PUBLIC.V$ARCHIVED_LOG SELECT
PUBLIC.V$DATABASE SELECT
PUBLIC.V$INSTANCE SELECT
PUBLIC.V$LOGMNR_CONTENTS SELECT
PUBLIC.V$NLS_PARAMETERS SELECT
PUBLIC.V$PARAMETER SELECT
PUBLIC.V$TRANSACTION SELECT

Object Name Object Privilege
SYS.DBA_LOG_GROUPS SELECT
SYS.DBA_LOG_GROUP_COLUMNS SELECT
SYS.DBMS_FLASHBACK EXECUTE
SYS.DBMS_LOGMNR EXECUTE
SYS.DBMS_LOGMNR_D EXECUTE
Step 7. Configuring Oracle Minimal Global Supplemental Logging

PowerExchange requires Oracle to use minimal global supplemental logging for Oracle LogMiner to properly
handle chained rows.
To enable minimal global supplemental logging, log in to the Oracle database and execute the following SQL
statement, which is included in the oracapt.sql and oracapt_rac.sql configuration files:
ALTER DATABASE ADD SUPPLEMENTAL LOG DATA;
COMMIT;
If you do not know whether minimal global supplemental logging has been enabled for your database, you can still
execute this ALTER statement. The statement has no effect if minimal supplemental logging is active.
Note: You must also define a supplemental log group for each Oracle source table. When you register an Oracle
source table in the PowerExchange Navigator, PowerExchange generates DDL for adding a supplemental log
group for the table. Oracle supplemental log groups cause Oracle to log full before- and after-images of the data
that changed. PowerExchange requires these images to properly process changes.
Step 8. Create a Table Space for Oracle LogMiner Use (Optional)

Create a table space exclusively for Oracle LogMiner use. This step is necessary only if you have not previously
configured LogMiner for use with other Oracle features such as logical standby databases, Oracle Streams, or
native Oracle change capture processes.
This step prevents the SYSTEM table space (in Oracle 9i) or SYSAUX table space (in Oracle 10g or later) from
becoming full and causing service problems during PowerExchange CDC.
To create the LogMiner table space, use the DDL in the PowerExchange oracapt.sql or oracapt_rac.sql file that is
supplied for this purpose.
1. To create the table space, issue the following DDL:

CREATE TABLESPACE "LOGMNRTS"
NOLOGGING
DATAFILE
'/oracle_path/datafilename.ora' SIZE
50M REUSE AUTOEXTEND
ON NEXT 10M MAXSIZE 100M EXTENT MANAGEMENT LOCAL;
Specify NOLOGGING if you use Oracle LogMiner only for PowerExchange CDC and an occasional query. Change
NOLOGGING to LOGGING if you use any of the following Oracle features: logical standby databases, Oracle
Streams, or native Oracle change capture processes.

For the DATAFILE value, specify a file name based on your local Oracle database file naming standards for
the data files that comprise this table space.
2. Enter the following command:
EXECUTE SYS.DBMS_LOGMNR_D.SET_TABLESPACE('LOGMNRTS');
If this statement fails with the ORA_01353 message, see the comments in oracapt.sql for more information.
3. To recompile the SYS.DBMS_LOGMNR_D package, enter the following command:
ALTER PACKAGE SYS.DBMS_LOGMNR_D COMPILE BODY;
Tip: LogMiner opens a number of cursors internally to handle its processing. When you configure LogMiner
for the first time, you might receive messages that state “number of open cursors exceeded.” You can
increase the maximum number of open cursors to handle the extra LogMiner processing.
Step 9. Copy the Oracle Catalog to the Archived Logs

PowerExchange CDC requires a copy of the Oracle online catalog in the Oracle archived redo logs to determine
the point from which to restart change data extractions.
PowerExchange reads the last catalog copy in the archived logs, even if you specified ONLINECAT=Y in the
ORCL CAPI_CONNECTION statement. You should copy the catalog on a routine basis to minimize CDC restart
times.
To copy the catalog, issue the following command in an SQL*Plus session:

begin
SYS.DBMS_LOGMNR_D.BUILD(
options => sys.dbms_logmnr_d.store_in_redo_logs);
end;
/
Tip: Periodically, PowerExchange requests Oracle to recopy the catalog to the Oracle archived redo logs. To
control how often Oracle copies the catalog and the time period within which the copy operation can occur, set the
CATBEGIN, CATEND, and CATINT parameters in the ORCL CAPI_CONNECTION statement of the dbmover.cfg
file.
Configuration in an Oracle RAC Environment

If you use Oracle 10g Release 2 or later, PowerExchange can process change data for database instances in a
real application cluster (RAC) environment. Certain Oracle patches might be required. For more information, see
Knowledge Base (KB) item 102503.
The Oracle instance from which you run PowerExchange CDC must be able to access the Oracle archived redo
logs for all Oracle instances in the RAC for which you want to capture change data.
In the init.ora file for each of these Oracle instances, define the LOG_ARCHIVE_DEST_1 parameter to point to the
directory in which you want Oracle to create the archived logs.
Note: PowerExchange uses Oracle LogMiner to read change data from the archived logs. If you use an archived
log destination other than the LOG_ARCHIVE_DEST_1 path and LogMiner processing lags behind, problems
might occur. In this situation, LogMiner starts reading change data from the archived logs in the
LOG_ARCHIVE_DEST_1 directory. If these archived logs are inaccessible from the machine with the Oracle
instance to which you are connected, the LogMiner session might fail.
Additional tasks for ensuring access to archived redo logs vary by operating system.
On Windows, you must set up an Oracle flash recovery area on the shared file system that contains all of the table
data for the RAC. For each Oracle instance in the RAC, set the LOG_ARCHIVE_DEST_1 parameter to point to
that recovery area.

On Linux and UNIX, you can use any of the following methods:
¨ Set up an Oracle flash recovery area in the same manner as for Windows.
¨ Store all archived redo logs on shared storage.

¨ Set up Network File System (NFS) access to the archive logs.
If you use shared storage or NFS access, the Oracle instance from which you run CDC must access the archived
logs of the other RAC member instances. This access uses the mount points that match the archive log directories
defined for those member instances. For example, assume that ORA2 is an Oracle instance in a RAC, which has a
LOG_ARCHIVE_DEST_1 parameter that points to the following archive log directory:
/ora/arch2/
ORA1 is the Oracle instance that runs CDC. The mount point that the ORA1 machine must use to access the
ORA2 archive logs is also /ora/arch2/.
Also, all of the Oracle instances in the RAC that participate in CDC must have access to the Oracle online redo
logs. Usually, these redo logs reside on shared storage.
PowerExchange Configuration for Oracle LogMiner CDC

The tasks that you perform to configure PowerExchange for CDC depend whether you want to use the
PowerExchange Logger for Linux, UNIX, and Windows and the extraction mode that you plan to use.
Configuring Oracle LogMiner CDC without the PowerExchange Logger

and Windows, complete the following tasks to configure PowerExchange for Oracle LogMiner CDC:
1. When you configure the dbmover.cfg file on the Oracle source machine, include the following statements:
¨ CAPT_PATH
¨ CAPT_XTRA
¨ ORACLEID
¨ ORCL CAPI_CONNECTION
¨ UOWC CAPI_CONNECTION
For more information, see the PowerExchange Reference Manual.

2. In the PowerExchange Navigator, create a capture registration for each Oracle source table.
create new ones.
You must enter a name in the Supplemental Log Group Name field.
unless you have a specific reason not to do so. This practice prevents having to edit the capture registrations
later if you decide to use the PowerExchange Logger. You might want to set the Condense option to None if
you plan to run both real-time and continuous extractions against tables defined by the same capture
registrations and do not want the PowerExchange Logger to capture change data for some registered tables.
The PowerExchange Navigator generates a corresponding extraction map and the DDL for creating a
supplemental log group. If you selected the Execute DDL now option, PowerExchange executes the DDL for

creating a supplemental log group when you click Finish. If you did not select this option, you must execute
the DDL prior to starting extraction processing.
RELATED TOPICS:
¨ “Customizing dbmover.cfg for Oracle LogMiner CDC” on page 90
Configuring Oracle LogMiner CDC with the PowerExchange Logger

If you plan to use the PowerExchange Logger for Linux, UNIX, and Windows and run extractions in batch or
continuous extraction mode, complete the following tasks to configure PowerExchange for Oracle LogMiner CDC:
1. When you configure the dbmover.cfg file used to access the source tables, include the following statements:
¨ CAPT_PATH
¨ CAPT_XTRA
¨ ORACLEID
¨ ORCL CAPI_CONNECTION
¨ UOWC CAPI_CONNECTION
For more information, see the PowerExchange Reference Manual.

3. Start the PowerExchange Listener on the source machine.
4. Customize the dbmover.cfg files on the Windows machine where the PowerExchange Navigator runs and on
the PowerCenter Integration Service machine, if these machines are separate from the Oracle source
machine.
In each of these dbmover.cfg files, you must specify a NODE statement that points to the machine that
contains the Oracle source tables. On the Windows machine, you must also specify an ORACLEID statement.
5. In the PowerExchange Navigator, create a capture registration for each Oracle source table.
create new ones.
You must select Part in the Condense list, and enter a name in the Supplemental Log Group Name field.
You can also set the Status option to Active, or wait until after you materialize the target tables.
The PowerExchange Navigator generates a corresponding extraction map and the DDL for creating a
supplemental log group. If you selected the Execute DDL now option, PowerExchange executes the DDL for
creating a supplemental log group when you click Finish. If you did not select this option, you must execute
the DDL prior to starting extraction processing.
6. In the PowerExchange Navigator, perform a database row test on the extraction maps to verify that
PowerExchange can access the source data.
7. After stopping updates to the source tables, materialize the target tables.
9. Allow changes to be written to the source tables.
mode.
PowerExchange Configuration for Oracle LogMiner CDC 89

RELATED TOPICS:
¨ “Customizing dbmover.cfg for Oracle LogMiner CDC” on page 90
Customizing dbmover.cfg for Oracle LogMiner CDC

In the dbmover.cfg configuration file, include the statements that are required for Oracle LogMiner CDC and any
optional statements that you want to use.
The following statements are required for Oracle CDC with Oracle LogMiner:
CAPT_PATH
Path to the local directory where the CCT file and CDCT file reside. The CCT file contains capture
registrations. The CDCT file contains information about PowerExchange Logger log files.
CAPT_XTRA
Path to the local directory where extraction maps reside.
ORACLEID
Oracle source instance, database, and connection information.
ORCL CAPI_CONNECTION
A named set of parameters that the CAPI uses to connect to the change stream and control extraction
processing for Oracle sources.
UOWC CAPI_CONNECTION
A named set of parameters for the UOW Cleanser. The CAPINAME parameter in the UOWC
CAPI_CONNECTION points to an ORCL CAPI_CONNECTION.
CAPX CAPI_CONNECTION (required for continuous extraction only)
If you plan to use the PowerExchange Logger and continuous extraction mode, you must also define a CAPX
CAPI_CONNECTION statement.
Define the CAPI_CONNECTION statements in the dbmover.cfg file that is on the system where the Oracle capture
registrations are stored. This location corresponds to the Location node that you specify when defining a
registration group. Usually, this location is where the source database resides.
Additionally, Informatica recommends including the LOGPATH and TRACING statements to make finding
messages easier. The LOGPATH statement defines a directory specifically for PowerExchange message log files,
and the TRACING statement enables PowerExchange to create an alternative set of message log files for each
PowerExchange process.
For more information about all dbmover.cfg statements, see the PowerExchange Reference Manual.
Example Oracle LogMiner CDC Statements

The following statements are typical of those included in a dmover.cfg for Oracle LogMiner CDC:
LOGPATH=/pwx/logs
TRACING=/PFX=PWXLOG,RECLEN=255,FILENUM=3,APPEND=Y,FLUSH=99)
CAPT_XTRA=/pwx/capture/vnnn/camaps
CAPT_PATH=/aus/pwx/capture/vnnn
ORACLEID=(FOX123,FO920DTL)
CAPI_SRC_DFLT=(ORA,CAPIUOWC)

CAPI_CONN_NAME=CAPIUOWC
/*
/* CAPI connection statements
/*
/* Both UOWC and ORCL CAPI_CONNECITON statements are required for Oracle CDC.
CAPI_CONNECTION=(NAME=CAPIORA
,DLLTRACE=ORA2
,TYPE=(ORCL
,ARRAYSIZE=1000
,BYPASSUF=Y
,CATBEGIN=00:01
,CATEND=23:59
,CATINT=1440
,ORACOLL=FOX123
,SELRETRY=0))
CAPI_CONNECTION=(NAME=CAPIUOWC
,TYPE=(UOWC
,CAPINAME=CAPIORA
,MEMCACHE=50000
,RSTRADV=1800))
/* Additional CAPX CAPI_CONNECTION statement is required for continuous extraction mode.
CAPI_CONNECTION=(NAME=CAPXORA
,TYPE=(CAPX
,DFLTINST=FOX920))
ORACLEID Statement
The ORACLEID statement specifies the Oracle source database and connection information for PowerExchange
CDC with Oracle LogMiner.
Data Sources: Oracle CDC sources

Required: Yes for Oracle LogMiner
CDC
Syntax:
ORACLEID=(
collection_id,
oracle_db,
[source_connect_string,]
[capture_connect_string,]
)
Parameters:
Enter the following positional parameters:
capture_connect_string
Optional. Oracle connection string, defined in TNS, that the PowerExchange Logger uses to connect to the
Oracle database with the source tables for Oracle LogMiner CDC. This connection string must be specified in
the Oracle Client tnsnames.ora file that is used for connection to the Oracle source database.
If this value is null, the value of the ORACLE_SID environment variable is used by default and the
PowerExchange Logger does not use Oracle SQL*Net for connection. If the ORACLE_SID environment
variable is not defined, the default Oracle database is used, if defined.
For Oracle LogMiner CDC only, if you have multiple Oracle databases and capture changes from a database
other than the default database, you must specify both the source_connect_string and capture_connect_string
parameters.
Tip: If possible, bypass the use of SQL*Net to improve PowerExchange Logger performance, even if the
PowerExchange Logger is running on the same machine as the Oracle source database. Set the following

environment variables, whenever possible, to enable connection to the appropriate Oracle database without
using the capture_connect_string parameter and SQL*Net:
¨ ORACLE_HOME
¨ ORACLE_SID
¨ PATH
¨ On a Linux or UNIX operating system, one of the following variables: LD_LIBRARY_PATH, LIBPATH, or
SHLIB_PATH
collection_id
Required. User-defined identifier for this ORACLEID statement. This value must match the ORACOLL
parameter value in the ORCL CAPI_CONNECTION statement, the collection ID in the registration group
defined for the source tables, and the DBID value in the PowerExchange Logger pwxccl.cfg file. Maximum
length is eight characters.
oracle_db
Required. Name of the Oracle database that contains the source tables you registered for change data
capture.
source_connect_string
Optional. Oracle connection string, defined in TNS, that is used to connect to the Oracle database that
contains the source tables. This connection string must be defined in the Oracle Client tnsnames.ora file on
the machine with the source database.
For Oracle LogMiner CDC, the source connection string is used only for PowerExchange Navigator access to
the Oracle source database. Enter this parameter in the dbmover.cfg file on the machine from which the
PowerExchange Listener retrieves data for PowerExchange Navigator requests. If you plan to run a database
row test on extraction maps for the source tables, also specify the capture_connect_string parameter.
Note: The source connection string is not used to transfer change data.
If this value is null, the value of the ORACLE_SID environment variable is used by default. If the
ORACLE_SID environment variable is not defined, the default Oracle database is used, if defined.
Usage Notes:
PowerExchange requires an ORACLEID statement for each Oracle database for which you want to capture and
extract change data. You can specify a maximum of 20 ORACLEID statements in a single dbmover.cfg file.
Specify the ORACLEID statement in the dbmover.cfg file on the machine where the PowerExchange Logger runs,
or if you plan to perform Oracle LogMiner CDC without the PowerExchange Logger, on the machine where your
PowerExchange extractions run.
ORCL CAPI_CONNECTION Statement

The ORCL CAPI_CONNECTION statement specifies the Consumer API (CAPI) parameters needed for Oracle
CDC sources that use Oracle LogMiner.
Data Sources: Oracle sources

Related UOWC
Statements: CAPI_CONNECTION
Required: Yes for Oracle LogMiner
CDC
Syntax:
CAPI_CONNECTION=(
NAME=name,

[TRACE=trace,]
TYPE=(ORCL,
[ARRAYSIZE=array_size,]
[BYPASSUF={N|Y},]
[CATBEGIN=hh:mm,]
[CATEND=hh:mm,]
[CATINT=minutes,]
[COMMITINT=minutes,]
[GENRLOCK={N|Y},]
[IGNUFMSG={N|Y},]
[LOGDEST=logdest_id,]
[LGTHREAD=instance_number,]
[ONLINECAT={N|Y},]
ORACOLL=collection_id,
[SELRETRY=retry_number,]
[SNGLINST={N|Y}]
)
)
Parameters:
DLLTRACE=trace_id
NAME=name
TRACE=trace
TYPE=(ORCL, ... )
Required. Type of CAPI_CONNECTION statement. For Oracle CDC sources that use LogMiner, this value
must be ORCL.
ARRAYSIZE=array_size
Optional. Size, in number of rows, of the prefetch array that PowerExchange uses to read the Oracle
redo logs. A value of less than 100 can degrade Oracle CDC performance.
Note: A value of 0 disables prefetch. Specify 0 only at the direction of Informatica Global Customer
Support.
Default is 100.
BYPASSUF={N|Y}
Optional. Controls whether PowerExchange ends abnormally or issues a warning message when an
unformatted log record is returned from Oracle LogMiner.
LogMiner returns unformatted log records when Global Temporary Tables are updated, or when
ONLINECAT=Y is specified and the log data that is being read is inconsistent with the catalog.

¨ N. PowerExchange ends with an error whenever it receives an unformatted log record from Oracle
LogMiner.
¨ Y. PowerExchange writes a warning message to the PowerExchange message log that warns that
unformatted log data has been found, and then continues processing. Depending on the amount of
unformatted log data, many warning messages might be written. You can specify Y for the
IGNUFMSG parameter to suppress these warning messages.
Default is N.
Tip: Specify Y if the Oracle instance contains Global Temporary tables. Otherwise, do not include the
BYPASSUF parameter.
CATBEGIN=hh:mm
Optional. Earliest time of day, in 24-hour clock format, at which PowerExchange requests Oracle to write
a copy of the Oracle catalog to the redo logs.
If you specify a value for the CATBEGIN parameter, you must also specify a value for the CATEND
parameter.
Default is 00:00.
CATEND=hh:mm
Optional. Latest time of day, in 24-hour clock format, at which PowerExchange requests Oracle to write a
copy of the Oracle catalog to the redo logs.
If you specify a value for the CATEND parameter, you must also specify a value for the CATBEGIN
parameter.
Default is 24:00.
CATINT=minutes
Optional. Time interval, in minutes, between requests to copy the Oracle catalog to the redo logs.
If this interval elapses but the time is outside of the time period specified in the CATBEGIN and CATEND
parameters, PowerExchange does not request Oracle to take a copy of the Oracle catalog. Instead,
PowerExchange waits until the time specified for the CATBEGIN parameter to request a catalog copy.
Default is 1440.
COMMITINT=minutes
Optional. Time interval, in minutes, between the SQL COMMIT operations issued by PowerExchange to
commit the transactions automatically generated by the Oracle LogMiner session.
Although PowerExchange does not update data in user tables while reading change data from the redo
logs, the Oracle LogMiner interface automatically generates transactions for the LogMiner sessions that
PowerExchange initiates. Oracle leaves these transactions open, or in-flight, until the LogMiner session
ends.
To be able to restart change data extraction operations efficiently, PowerExchange must occasionally
issue SQL COMMIT operations to end these in-flight transactions. Otherwise, the restart of all future real-
time extraction operations might be impacted because PowerExchange always begins reading change
data at the beginning of the oldest in-flight UOW.

Default is 5.
GENRLOCK={N|Y}
Optional. Controls whether PowerExchange generates a safe restart point for requests for restart points
that match the current end-of-log (EOL).
¨ N. PowerExchange generates restart points that match the current EOL, ignoring any in-flight
transactions for the source tables.
¨ Y. PowerExchange generates safe restart points for source tables.
A safe restart point for a source table is a point in the change stream that does not skip any in-flight
UOWs for that table. To generate a safe restart point for a source table, PowerExchange obtains an
exclusive lock on the table to stop further changes. PowerExchange then searches the Oracle catalog for
the point in the change stream that matches the earliest active transaction for the table and uses this
point as the restart point. If no in-flight UOWs exist for a table, PowerExchange uses the current EOL.
PowerExchange releases the lock on the source table after the restart point generation process
completes, which allows new changes to the table to occur.
PowerExchange generates restart tokens that match the current EOL in the following situations:
¨ The PowerExchange Logger for Linux, UNIX, and Windows is cold started and the pwxccl.cfg
configuration file does not specify the SEQUENCE_TOKEN and RESTART_TOKEN parameters.
PowerExchange obtains locks for all tables represented by capture registrations selected for
processing by the PowerExchange Logger.
¨ The restart token file for a CDC session specifies the CURRENT_RESTART option on the RESTART1
and RESTART2 special override statements.
PowerExchange obtains locks only for the tables in the CDC session to which the special override
statements apply.
¨ A database row test in the PowerExchange Navigator that uses the SELECT CURRENT_RESTART
SQL statement.
PowerExchange obtains a lock for the table represented by capture registration associated with the
extraction map used in the database row test.
¨ A DTLUAPPL utility operation that uses the RSTTKN GENERATE option.
PowerExchange obtains a lock for the table represented by the capture registration specified in the
utility control statements.
Default is N.
IGNUFMSG={N|Y}
Optional. Controls whether PowerExchange writes warning messages to the PowerExchange message
log file for unformatted data records.
¨ N. PowerExchange does not write any warning messages.
¨ Y. PowerExchange writes warning messages.
Default is N.

LOGDEST=logdest_id
Optional. For RAC environments, the numeric identifier for the archive log destination that you want to
force PowerExchange to use. This archive log destination must be local to the Oracle instance that
PowerExchange is using.
For example, to use archived logs from the destination set by the LOG_ARCHIVE_DEST_3 parameter in
the init.ora file, specify LOGDEST=3.
The SNGINST parameter affects how PowerExchange uses the archive log destination and the Oracle
instance specified by LOGDEST and LGTHREAD.
If you specify Y for the ONLINECAT parameter, PowerExchange validates and then ignores the
LOGDEST and LGTHREAD parameters.
LGTHREAD=instance_number
Optional. For RAC environments, the numeric instance number for the Oracle instance that
PowerExchange uses to identify the archived redo logs to process.
The SNGINST parameter affects how PowerExchange uses the archive log destination and the Oracle
instance specified by LOGDEST and LGTHREAD.
If you specify Y for the ONLINECAT parameter, PowerExchange validates and then ignores the
LOGDEST and LGTHREAD parameters.
ONLINECAT={N|Y}
Optional. Controls whether PowerExchange directs Oracle LogMiner to use the Oracle online catalog or
the copy of the catalog in the redo logs to format log data for CDC.
¨ N. Oracle LogMiner uses the copy of the catalog from the archived redo logs and PowerExchange
tracks schema changes to ensure that data loss does not occur.
¨ Y. Oracle LogMiner uses the online catalog and PowerExchange cannot track schema changes.
When PowerExchange is configured to use the online catalog for formatting log data, it still uses catalog
copies to determine the restart point for change data extraction operations. Therefore, you must copy the
online catalog to the Oracle redo logs on a regular basis.
Change data extraction operations generally initialize faster when PowerExchange is configured to create
LogMiner sessions with the online catalog instead of a catalog copy. However, when LogMiner uses the
online catalog, it does not track DDL changes, and cannot format log records for tables that have schema
changes.
If LogMiner uses the online catalog and you make schema changes made while LogMiner is reading log
data, LogMiner passes unformatted log records for subsequent changes to PowerExchange. If you
specify N for the BYPASSUF parameter, or allow it to default, PowerExchange fails the extraction request
after Oracle passes the first unformatted record. Otherwise, PowerExchange skips the unformatted
record and continues processing, which results in change data loss. Therefore, specify N for the
ONLINECAT parameter, or allow it to default, if you have the following requirements:
¨ You specify Y for the BYPASSUF parameter and need to change the schema of tables registered for
capture while change data extraction operations are running.
¨ You need to start an extraction from a point in the Oracle redo logs that contains table data that was
captured under a previous schema.

Default is N.
ORACOLL=collection_id
Required. Oracle collection identifier, which must match the value specified in the ORACLEID statement.
SELRETRY=retry_number
Optional. Number of times that PowerExchange immediately loops back to the Oracle LogMiner call
before implementing a graduated-scale wait loop.
After the call to LogMiner has been retried the specified number of times, PowerExchange implements a
wait interval between each subsequent retry. The wait interval begins at one millisecond and gradually
increases to one second. When LogMiner returns data, the wait interval is reset to 0, and the process
begins again for the next call to LogMiner.
If you specify a non-zero value, PowerExchange uses non-blocking SQL to ensure that a user request to
shut down a extraction session is processed in a timely manner.
If you specify 0, PowerExchange does not use non-blocking SQL. This setting improves CPU
consumption but can prolong extraction session shutdown. On quiescent Oracle instances,
PowerExchange does not honor a shutdown request until log data is returned from Oracle. On Oracle
instances where update activity is occurring, shutdown behavior does not noticeably change.
Default is 1000.
SNGLINST={N|Y}
Optional. In RAC environments, controls whether PowerExchange uses only the archived redo logs from
a specific Oracle instance and archive log destination.
¨ N. PowerExchange uses the specified Oracle instance to search for archived redo logs that contain
copies of the Oracle catalog. After PowerExchange passes these logs to an Oracle LogMiner session,
LogMiner determines the other archived redo logs to read.
¨ Y. PowerExchange uses only the archive log destination and Oracle instance that you specify in
LOGDEST and LGTHREAD parameters to read archived redo logs. LogMiner does not read any other
archived redo logs. After PowerExchange processes the logs from the specified location, the change
data extraction operation ends.
If you specify Y, you must also specify the LOGDEST and LGTHREAD parameters to identify the archive
log destination and Oracle instance to use. For all remaining Oracle instances in the RAC, you must run
separate change data extraction processes and then determine how to properly merge the change data
so that you can apply it to targets.
Default is N.
Oracle Catalog Parameters in the ORCL CAPI_CONNECTION Statement

The CATINT, CATBEGIN and CATEND parameters in the ORCL CAP_CONNECTION statement can significantly
affect PowerExchange performance. These parameters control the frequency with which the Oracle catalog is
copied to the Oracle redo logs and the time period within which the copy operation can occur. When you restart
PowerExchange extraction processing, PowerExchange directs Oracle LogMiner to begin reading change data
from the redo logs starting from the SCN of the last Oracle catalog copy that was written to the logs prior to the
end of the previous extraction session.

To configure the CATINT, CATBEGIN, and CATEND parameters, try various settings until you find a combination
that provides for efficient restart processing. The default frequency of once a day might not be sufficient if you
have a high volume of transaction activity.
The following examples demonstrate how copying the Oracle catalog multiple times can affect the amount of
change data that is reread from the archived redo logs when PowerExchange extraction processing is restarted.
Example 1
Assume that the Oracle catalog was initially copied to the Oracle redo logs at SCN 10 and another copy has not
yet been written to the logs. Change data was logged starting at SCN 40 and ending at SCN 60. A
PowerExchange extraction session extracted these changes before ending at SCN 100. Since the extraction
session ended, additional changes have been logged starting at SCN 160.
When you restart PowerExchange extraction processing, LogMiner must begin reading change data from the initial
catalog copy at SCN 10 because it is the latest catalog copy prior to the session end at SCN 100. As a result,
PowerExchange reprocesses the data between SCN 10 and SCN 100, before continuing to the new change data
that begins at SCN 160. This reprocessing of data impacts PowerExchange performance.
Example 2
Assume that the Oracle catalog was copied to the Oracle redo logs twice: at SCN 10 and at SCN 80. Change data
was logged starting at SCN 40 and ending at SCN 60. A PowerExchange extraction session extracted these
changes before ending at SCN 100. Since the extraction session ended, additional changes have been logged
starting at SCN 160.
When you restart PowerExchange extraction processing, LogMiner begins reading change data from the second
catalog copy at SCN 80 because it is the latest catalog copy prior to the session end at SCN 100. As a result,
PowerExchange reprocesses only the data between SCN 80 and SCN 100, before continuing to the new change
data that begins at SCN 160. With multiple catalog copies, PowerExchange needs to reprocess less change data.

UOWC CAPI_CONNECTION Statement
The UOWC CAPI_CONNECTION statement specifies the Consumer API (CAPI) parameters needed for the UOW
Cleanser.
In the change stream for some data sources, changes from multiple UOWs are intermingled. The UOW Cleanser
reconstructs the intermingled changes read from the change stream into complete UOWs in chronological order
based on end time.
Data Sources: DB2 for i5/OS

Oracle LogMiner CDC
z/OS CDC
Related AS4J
Statements: CAPI_CONNECTION for
i5/OS
ORCL
CAPI_CONNECTION for
Oracle
LRAP
CAPI_CONNECTION for
z/OS
Required: Yes for the noted data
sources
Syntax:
CAPI_CONNECTION=(
NAME=name,
[TRACE=trace,]
TYPE=(UOWC,
CAPINAME=name,
[BLKSIZE=block_size,]
[DATACLAS=data_class,]
[RSTRADV=seconds,]
[SPACEPRI=primary_space,]
[SPACETYPE={BLK|TRK|CYL},]
[STORCLAS=storage_class,]
[UNIT=unit]
)
)
Parameters:
DLLTRACE=trace_id
NAME=name
TRACE=trace
TYPE=(UOWC, ... )
Required. Type of CAPI_CONNECTION statement. For the UOW Cleanser, this value must be UOWC.

BLKSIZE=block_size
Optional. Block size, in bytes, for the sequential UOW spill files that the UOW Cleanser creates when the
memory cache cannot hold all changes for a UOW.
Valid values and defaults vary by platform:
¨ For Oracle LogMiner CDC sources, enter a value from 8 through 65535. Default is 32768.
¨ For i5/OS CDC sources, enter a value from 8 through 32760. Default is 32760.
¨ For z/OS CDC sources, enter a value from 8 through 32760. Default is 18452.
CAPINAME=name
Required. Value from the NAME parameter in the related source-specific CAPI_CONNECTION statement.
The source-specific CAPI_CONNECTION is one of the following statement types:
¨ AS4J CAPI_CONNECTION statement for i5/OS CDC sources
¨ LRAP CAPI_CONNECTION statement for z/OS CDC sources
¨ ORCL CAPI_CONNECTION statement for Oracle LogMiner CDC sources
DATACLAS=data_class
Optional. On z/OS, the SMS data class that the UOW Cleanser uses when allocating the sequential UOW
spill files. If you do not specify this parameter, the SMS ACS routines can assign the data class.
MEMCACHE=cache_size
Optional. Memory cache size, in kilobytes, that PowerExchange allocates to reconstruct complete UOWs.
For each extraction session, PowerExchange keeps all changes for each UOW in the memory cache until
it processes the end-UOW record. If the memory cache is too small to hold all of the changes in a UOW,
PowerExchange spills the changes to a sequential files on disk, called UOW spill files.
Each UOW spill file contains one UOW. A UOW might require multiple UOW spill files to hold all of the
changes for that UOW. If the change stream contains multiple large UOWs and the memory cache is
insufficient, PowerExchange might create numerous UOW spill files.
PowerExchange processes the change stream more efficiently if it does not need to use UOW spill files.
In addition to degrading extraction performance, large numbers of UOW spill files can cause a disk space
shortage.
Important: If the change stream contains only small UOWs, the default value might be sufficient.
However, the default value is often too small to eliminate UOW spill files. Informatica recommends that
so you specify a larger value.
The location in which PowerExchange allocates the UOW spill files varies by operating system, as
follows:
¨ For i5/OS, PowerExchange uses CRTPF command to create a physical file for UOW spill files.
PowerExchange creates the UOW spill file names by using the C/C++ tmpnam() function.
¨ For Linux and UNIX, PowerExchange uses the current directory by default for UOW spill files. To use
a different directory, specify the TMPDIR environment variable.
PowerExchange creates the UOW spill file names by using the operating system tempnam function
with a prefix of dtlq.
Note: The UOW spill files are temporary files that are deleted when PowerExchange closes them.
They are not visible in the directory while open.

¨ For Windows, PowerExchange uses the current directory by default for UOW spill files. To use a
different directory, specify the TMP environment variable.
PowerExchange creates the UOW spill file names by using the Windows _tempnam function with a
prefix of dtlq.
¨ For z/OS, PowerExchange uses dynamic allocation to allocate temporary data sets for the UOW spill
files. Generally, SMS controls the location of temporary data sets. If you do not use SMS to control
temporary data sets, the UNIT parameter controls the location for the UOW spill files.
Because PowerExchange allocates temporary data sets for the UOW spill files, z/OS assigns these
files system-generated data set names, which begin with SYSyyddd.Thhmmss.RA000.jobname.
Warning: Because PowerExchange allocates the cache size for each extraction operation, use caution
when coding large values for MEMCACHE. Otherwise, many concurrent extraction sessions might cause
memory constraints.
Default is 1024, or 1 MB.
RSTRADV=nnnnn
restart point.
each UOW processed.
SPACEPRI=primary_space
Optional. On z/OS, the primary space value that the UOW Cleanser uses to allocate UOW spill files. The
UOW Cleanser does not use secondary space. Instead, when a spill file becomes full, the UOW Cleanser
allocates another spill file of the same size. The SPACETYP parameter specifies the space units for this
value. Default is 50 cylinders.
SMS ACS routines can override the UOW spill file size.
Default is 50 cylinders.
Note: On i5/OS, the UOW Cleanser allocates UOW spill files as physical files with SIZE(*NOMAX), which
means that the maximum spill file size is controlled by the system maximum file size. On Linux, UNIX,
and Windows, PowerExchange allocates UOW spill files as temporary files that are 2 GB in size.

SPACETYPE={BLK|TRK|CYL}
Optional. On z/OS, the type of space units that the UOW Cleanser uses to allocate UOW spill files.
¨ BLK. Use blocks.
¨ CYL. Use cylinders.
¨ TRK. Use tracks.
Default is BLK.
STORCLAS=storage_class
Optional. On z/OS, the SMS storage class name that the UOW Cleanser uses to allocate UOW spill files.
UNIT=unit
Optional. On z/OS, the generic or esoteric unit name that the UOW Cleanser uses to allocate UOW spill
files.
Management of Oracle LogMiner CDC

You might need to stop CDC for source tables occasionally, for example, to change the table definitions.
Stopping Oracle LogMiner CDC

You might need to stop Oracle change data capture for a source table to perform troubleshooting or routine
maintenance tasks.
To stop change data capture, use one of the following methods:
¨ Open the capture registration for the source table, and change the Status value from Active to History.
Warning: A capture registration that has a status of History cannot be activated again. This method
permanently stops change data capture for a table based on a particular capture registration.
¨ Drop the supplemental log group by executing the following SQL:

ALTER TABLE schema.table_name DROP SUPPLEMENTAL LOG GROUP
After you drop the supplemental log group, Oracle stops recording full before- and after-images of data that
changed. If you reinstate the supplemental log group later, you should rematerialize the target database.
RELATED TOPICS:
¨ “Stopping PowerCenter CDC Sessions” on page 142
Changing a Source Table Definition Used in Oracle LogMiner CDC

Occasionally, you might need to change the definition of an Oracle source table that is registered for change data
capture. If your metadata changes affect the columns from which change data is captured, use this procedure to
enable PowerExchange to switch to the updated table definition, while preserving access to previously captured
data.

Perform this procedure whenever you add, alter, or drop columns for which change data is captured. You do not
need to perform this procedure if you are selectively capturing change data for a subset of columns and none of
the selected columns are affected by the table definition changes.
Tip: If you no longer need to capture change data from a column in a table, you can remove that column from the
extraction map without changing the capture registration. Change data for the column is still captured but is not
extracted.
To change a source table definition used in Oracle LogMiner CDC:

3. In the PowerExchange Navigator, open the original capture registration and set its status to History.
Note: PowerExchange does not capture change data based on capture registrations that have a status of
History or Inactive.
4. Use DDL to make the table changes.
5. Drop the supplemental log group for the table.
set its status to Active.
Also select the Execute DDL now option so that when you finish the capture registration, the PowerExchange
Navigator runs the DDL for creating a new supplemental log group.
PowerExchange uses the newly activated capture registration for change data capture.
8. If you use the PowerExchange Logger for Linux, UNIX, and Windows, restart the PowerExchange Logger
process so that it will begin using the new capture registration.
9. In PowerCenter Designer, import the altered source and target tables. Edit the mapping if necessary.
12. Restart extraction processing.
RELATED TOPICS:
Management of Oracle LogMiner CDC 103

Part IV: Change Data Extraction
¨ Introduction to Change Data Extraction, 105
¨ Extracting Change Data, 125
¨ Managing Change Data Extractions, 140
¨ Monitoring and Tuning Options, 148
104
CHAPTER 7
Introduction to Change Data

Extraction
¨ Change Data Extraction Overview, 105
¨ Extraction Modes, 106
¨ PowerExchange-Generated Columns in Extraction Maps, 106
¨ Restart Tokens and the Restart Token File, 109
¨ Recovery and Restart Processing for CDC Sessions, 111
¨ Group Source Processing in PowerExchange, 116
¨ Commit Processing with PWXPC, 118
¨ Offload Processing, 123
Change Data Extraction Overview

Use PowerExchange in conjunction with PWXPC and PowerCenter to extract captured change data and write it to
one or more targets. Review the topics in this chapter to learn key concepts about extraction processing so that
can configure CDC sessions to extract change data efficiently and to enable proper restart and recovery.
To extract changes captured by PowerExchange, import the metadata for the capture source into PowerCenter
Designer. Use one of the following methods:
¨ For nonrelational data sources, import the extraction map from PowerExchange.
¨ For relational data sources, you can import either the metadata from the database or the extraction map from
PowerExchange. If you import metadata from the database, you might need to modify the source definition in
Designer to add PowerExchange-defined CDC columns or to remove any columns that are not included in the
extraction map. If you import extraction maps, you do not need to manually add or remove these columns from
the PowerCenter source definition.
After you import the metadata, you can use the source definitions in PowerCenter to create mappings, sessions,
and workflows for extracting the change data from PowerExchange.
RELATED TOPICS:
¨ “PowerExchange-Generated Columns in Extraction Maps” on page 106
105
Extraction Modes
You can use different modes to extract change data captured by PowerExchange. The extraction mode is
determined by the PowerCenter connection type and certain PowerExchange CDC configuration parameters.
Some extraction modes are available only if you use PowerExchange Condense or the PowerExchange Logger for
Linux, UNIX, and Windows.
Depending on your extraction requirements, use one of the following extractions modes:
Real-time extraction mode
Continuously extracts change data directly from the PowerExchange Logger for MVS log files in near real
time. Extraction processing continues until the CDC session is stopped or interrupted.
To implement this mode, configure a PWX CDC Real Time application connection in PowerCenter for your
data source type.
Batch extraction mode
Extracts change data from PowerExchange Condense condense files on MVS that are closed at the time the
session runs. After processing the condense files, the CDC session ends.
To implement this mode, configure the following items:
¨ In PowerCenter, configure a PWX CDC Change application connection for your data source type.
¨ In the PowerExchange Navigator, set the Condense option to Part or Full in your capture registrations.
Continuous extraction mode.
Continuously extracts change data from open and closed PowerExchange Logger for Linux, UNIX, and
Windows log files in near real time.
To implement this mode, configure the following items:
¨ On the remote Linux, UNIX, or Windows system, configure the PowerExchange Logger for Linux, UNIX,
and Windows to log change data that was originally captured on MVS.
¨ In PowerCenter, configure a PWX CDC Real Time application connection for your data source type.
¨ In the PowerExchange Navigator, set the Condense option to Part in your capture registrations.
RELATED TOPICS:
¨ “Configuring PowerExchange to Capture Change Data on a Remote System” on page 162
¨ “Extracting Change Data Captured on a Remote System” on page 168
PowerExchange-Generated Columns in Extraction Maps

Besides the table columns defined in capture registrations, extraction maps include columns that PowerExchange
generates. These PowerExchange-generated columns contain CDC-related information, such as the change type
and timestamp.
When you import an extraction map in Designer, PWXPC includes the PowerExchange-generated columns in the
source definition.
When you perform a database row test on an extraction map, the PowerExchange Navigator displays the
PowerExchange-generated columns in the results. By default, the PowerExchange Navigator hides these columns
106 Chapter 7: Introduction to Change Data Extraction

from view when you open the extraction map. To display these columns, open the extraction map, right-click
anywhere within the Extract Definition window, and select Show Auto Generated Columns.
Note: By default, all columns except the DTL__columnname_CNT and DTL__columnname_IND columns are
selected in an extraction map. You must edit an extraction map to select these columns.
The following table describes the columns that PowerExchange generates for each change record:
Column Description Datatype Length
DTL__CAPXRESTART1 A binary value that represents the position of the end of the VARBIN 255
UOW for that change record followed by the position of the
change record itself.
The length of a sequence token varies by data source type,
except on z/OS where sequence tokens for all data source
types have the same length.
The value of DTL__CAPXRESTART1 is also known as the
sequence token, which when combined with the restart
token comprises the restart token pair.
A sequence token for a change record is a strictly
ascending and repeatable value.
DTL__CAPXRESTART2 A binary value that represents a position in the change VARBIN 255
stream that can be used to reconstruct the UOW state for
the change record, with the following exceptions:
- Microsoft SQL Server CDC. A binary value that contains
the DBID of the distribution database and the name of
the distribution server.
- Change data extracted from full condense files on z/OS
or i5/OS. A binary value that contains the instance name
from the registration group of the capture registration.
The length of a restart token varies by data source type. On
z/OS, restart tokens for all data source types have the
same length, except for change data extracted from full
condense files.
The value of DTL__CAPXRESTART2 is also known as the
restart token, which when combined with the sequence
token comprises the restart token pair.
DTL_CAPXRRN For DB2 on i5/OS only, the relative record number. DECIMAL 10
DTL__CAPXUOW A binary value that represents the position in the change VARBIN 255
stream of the start of the UOW for the change record.
DTL__CAPXUSER The user ID of the user that made the change to the data VARCHAR 255
source, with the following exceptions:
- DB2 for i5/OS. If you specify LIBASUSER=Y on the
AS4J CAPI_CONNECTION statement, the value is the
library and file name to which the change was made.
- DB2 for z/OS. If you do not specify UIDFMT on the LRAP
CAPI_CONNECTION, the value is the user ID of the
user that made the change. Otherwise, the UIDFMT
parameter determines the value.
- Microsoft SQL Server. The value is null because
Microsoft SQL Server does not record this information in
the distribution database.
- Oracle. The value might be null. If known, Oracle
provides the user ID.
PowerExchange-Generated Columns in Extraction Maps 107

Column Description Datatype Length
DTL__CAPXTIMESTAMP The timestamp for when the change was made to the data CHAR 20
source, as recorded by the source DBMS in the following
format:
YYYYMMDDhhmmssnnnnnn
Where:
- YYYYMMDD is the date in year (YYYY), month (MM),
and day (DD) format.
- hhmmssnnnnnn is the time in hours (hh), minutes (mm),
seconds (ss), and microseconds (nnnnnn) format.
Note: Oracle does not support microseconds in the
timestamp.
DTL__CAPXACTION A single character that indicates the type of change CHAR 1

operation. Valid values are:
- I. INSERT operation.
- D. DELETE operation.
- U. UPDATE operation.
DTL__CAPXCASDELIND For DB2 for z/OS sources only, a single character that CHAR 1
indicates whether DB2 has deleted the row because the
table specifies the ON DELETE CASCADE clause. Valid
values are:
- Y. Indicates that DB2 deleted this row because of a
cascade delete rule.
- N. Indicates that DB2 did not delete this row because of
a cascade delete rule.
DTL__BI_columnname For UPDATE operations, the value of the before image of Datatype of Length of the
the selected column in the change record. the source source
column column
DTL__CI_columnname For UPDATE operations, a single character that indicates CHAR 1

whether the selected column was changed. Valid values
are:
- Y. Indicates that the column changed.
- N. Indicates that the column did not changed.
- Null value. Indicates an INSERT or DELETE operation.
DTL__columnname_CNT Binary count column. PowerExchange generates this NUM32U 0

column for variable length columns of types VARCHAR and
VARBIN to determine the length of the column during
change data extraction processing.
Note: By default, binary count columns are not selected in
an extraction map. You must edit an extraction map to
select these columns.
DTL__columnname_IND Null indicator column. PowerExchange generates this BIN 1

column for nullable columns to indicate the nullable value
for the column.
Note: By default, null indicator columns are not selected in
an extraction map. You must edit an extraction map to
select these columns.

Restart Tokens and the Restart Token File
PowerExchange uses a pair of token values, called a restart token pair, to determine where to begin extracting
change data in the change stream for a CDC session. For a new CDC session, you should generate restart token
values that represent the point-in-time in the change stream where you materialized the targets. Each source in a
CDC session can have unique values for its restart token pair in the restart token file.
A restart token pair matches the position in the change stream for a change record and has the following parts:
Sequence token
For each change record that PowerExchange reads from the change stream, a binary value that represents
the change stream position of the end of the UOW for that change record followed by the position of the
change record itself, with the following exceptions:
¨ For Microsoft SQL Server CDC, a binary value that represents the position of the change record in the
distribution database.
¨ For change data extracted from full condense files on z/OS or i5/OS, a binary value that represents the full
condense file and the position of the change record in that file.
A sequence token for a change record is a strictly ascending and repeatable value. The length of a sequence
token varies by data source type, except on z/OS where sequence tokens for all data source types have the
same length.
Restart token
For each change record that PowerExchange reads from the change stream, a binary value that represents a
position in the change stream that can be used to reconstruct the UOW state for that record, with the following
exceptions:
¨ For Microsoft SQL Server CDC, a binary value that contains the DBID of the distribution database and the
name of the distribution server.
¨ For change data extracted from full condense files on z/OS and i5/OS, a binary value that contains the
instance name from the registration group for the capture registration.
In some cases, the restart token might contain the position of the oldest open UOW. An open UOW is a UOW
for which PowerExchange has read the beginning of the UOW from the change stream but has not yet read
the commit record, or end-UOW.
The length of a restart token varies by data source type. On z/OS, restart tokens for all data source types
have the same length, except for change data extracted from full condense files.
PowerExchange uses these restart token values to determine the point from which to start reading change data
from the change stream, with the following exceptions:
¨ For Microsoft SQL Server CDC, PowerExchange uses the sequence token value to determine the point from
which to start reading change data from that distribution database, and the restart token value to verify that the
distribution database is the same as the distribution database specified on the CAPI connection.
¨ For change data extracted from full condense files on z/OS or i5/OS, PowerExchange uses the sequence token
value to determine the point from which to start reading change data from the condense files, and the restart
token value to verify that the instance is the same as the instance recorded for the change record.
After determining the start point in the change stream for a CDC session, PowerExchange begins to read and
pass change data to PWXPC. PWXPC uses the sequence token value for each source in the CDC session to
determine the point at which to start providing the change data passed from PowerExchange to a specific source.
Restart Tokens and the Restart Token File 109

You should specify restart token values in the restart token file in the following situations:
¨ When creating a new CDC session, specify a restart token pair for each data source. Alternatively, you can use
the special override statement to specify a restart token pair for some or all data sources.
¨ When adding a data source to an existing CDC session, specify a restart token pair for the new source.
¨ If you need to override token values for a data source that is defined in an existing CDC session, specify the
override token values.
Generating Restart Tokens

Before you begin extracting change data, you must materialize the targets for the CDC session with data from the
data sources. Usually, to perform this task, you run a bulk data movement session. After you materialize the
targets and before you allow changes to be made to the data source again, you should generate restart tokens
that represent the point-in-time in the change stream when the materialization occurred.
PWXPC can generate restart tokens when it starts to extract change data for a CDC session. Additionally,
PowerExchange provides a number of methods to generate restart tokens. To generate restart tokens that match
the current end of the change stream, use one of the following methods:
¨ In the PWXPC restart token file for the CDC session, specify CURRENT_RESTART on the RESTART1 and
RESTART2 special override statements.
¨ In the PowerExchange Navigator, use the SELECT CURRENT_RESTART SQL statement when you perform a
database row test.
¨ Run the DTLUAPPL utility with the GENERATE RSTTKN option.
If you use the DTLUAPPL utility or the PowerExchange Navigator to generate restart tokens, edit the restart token
file that PWXPC uses to specify the token values before you start the CDC session.
Restart Token File

You can use the restart token file to provide restart tokens for a new CDC session, or for a source that you add to
an existing CDC session. You can also use the restart token file to override restart tokens for sources in an
existing CDC session.
Specify the name and location of the restart token file in the following attributes of the source PWX CDC
application connection:
¨ RestartToken File Folder
¨ RestartToken File Name
When you run a CDC session, PWXPC reads the restart token file in the folder specified in the RestartToken File
Folder attribute of the source CDC connection. If this folder does not exist and the RestartToken File Folder
attribute contains the default value of $PMRootDir/Restart, PWXPC creates this folder. PWXPC does not create
any other restart token folder name. PWXPC then verifies that the restart token file exists. If the file does not exist,
PWXPC uses the name specified in the RestartToken File Name attribute to create an empty restart token file.
PWXPC stores restart tokens for CDC sessions at the following locations:
¨ For relational targets, in a state table in the target database
¨ For nonrelational targets, in a state file on the PowerCenter Integration Service machine
When you restart a CDC session, PWXPC reads the restart tokens for each source in the CDC session from the
state table or file. PWXPC also reads the restart token file for the CDC session and overrides the restart tokens for
any sources that have token values included in the file.

Recovery and Restart Processing for CDC Sessions
If you select Resume from the last checkpoint for the Recovery Strategy attribute in a CDC session that
extracts change data from PowerExchange, PWXPC and PowerCenter provide recovery and restart processing for
that session. In the event of a session failure, the PowerCenter Integration Service recovers the session state of
operation, and PWXPC recovers the restart information.
PWXPC saves restart information for all sources in a CDC session. The restart information for CDC sessions,
which includes the restart tokens, originates from PowerExchange on the system from which the change data is
extracted. You can include both relational and nonrelational targets in a single CDC session. PWXPC uses one of
the following locations to store and retrieve restart information, based on the target type:
¨ Relational targets. Recovery state tables in the target databases. PWXPC, in conjunction with the
PowerCenter Integration Service, commits both the change data and the restart tokens for that data in the
same commit, which ensures that the applied data and the restart tokens are in-sync.
¨ Nonrelational targets. Recovery state file in the shared location on the PowerCenter Integration Service
machine. PWXPC, in conjunction with the PowerCenter Integration Service, writes the change data to the
target files and then writes the restart tokens to the recovery state file. As a result, duplicate data might be
applied to the targets when you restart failed CDC sessions.
The PowerCenter Integration Service saves the session state of operation and maintains target recovery tables.
The PowerCenter Integration Service stores the session state of operation in the shared location that is specified
in $PMStorageDir. The PowerCenter Integration Service saves relational target recovery information in the target
database.
When you run a CDC session that uses a resume recovery strategy, PWXPC writes the following message to the
session log to indicate that recovery is in effect:
PWXPC_12094 [INFO] [CDCRestart] Advanced GMD recovery in effect. Recovery is automatic.
When you recover or restart a CDC session, PWXPC uses the saved restart information to resume reading the
change data from the point of interruption. The PowerCenter Integration Service restores the session state of
operation, including the state of each source, target, and transformation. PWXPC, in conjunction with the
PowerCenter Integration Service, determines how much of the source data it needs to reprocess. PowerExchange
and PWXPC use the restart information to determine the correct point in the change stream from which to restart
extracting change data and then applying it to the targets.
If you run a session with resume recovery strategy and the session fails, do not change the mapping, the session,
or the state information before you restart the session. PowerCenter and PWXPC cannot guarantee recovery if
you make any of these changes.
Restriction: If any of the targets in the CDC session use the PowerCenter File Writer to write CDC data to flat
files, do not use a resume recovery strategy. Restart tokens for all targets in the CDC session, including relational
targets, will be compromised if a flat file target is in the same session. Data loss or duplication might occur.
PowerCenter Recovery Tables for Relational Targets

When the PowerCenter Integration Service runs a session that has a resume recovery strategy, it writes to
recovery tables on the target database system. When the PowerCenter Integration Service recovers the session, it
uses information in the recovery tables to determine where to begin loading data to target tables. PWXPC uses
information in the recovery tables to determine where to begin reading the change stream.
If you want the PowerCenter Integration Service to create the recovery tables, grant table creation privilege to the
database user name configured in the target database connection. Otherwise, you must create the recovery tables
manually.
Recovery and Restart Processing for CDC Sessions 111

For relational targets, the PowerCenter Integration Service creates the following recovery tables in the target
database:
¨ PM_RECOVERY. Contains target load information for the session run. The PowerCenter Integration Service
removes the information from this table after each successful session and initializes the information at the
beginning of subsequent sessions.
¨ PM_TGT_RUN_ID. Contains information the PowerCenter Integration Service uses to identify each target on
the database. The information remains in the table between session runs. If you manually create this table, you
must create a row and enter a value other than zero for LAST_TGT_RUN_ID to ensure that the session
recovers successfully.
¨ PM_REC_STATE. Contains state and restart information for CDC sessions. PWXPC stores the application
name and restart information for all sources in the CDC session. The PowerCenter Integration Service stores
any state information for the session. Unlike the session state information, restart information persists in this
table across successful sessions. The PowerCenter Integration Service updates it with each commit to the
target tables.
If you edit or drop the recovery tables before you recover a session, the PowerCenter Integration Service cannot
recover the session. Also, PWXPC cannot restart the CDC session from the point of interruption.
If you disable recovery, the PowerCenter Integration Service does not remove the recovery information from the
target database. Also, PWXPC no longer updates the restart information in the target database.
Recovery State Table

The recovery state table, PM_REC_STATE, contains state and CDC restart information for a CDC session. This
table resides in the same target database as the target tables.
The PowerCenter Integration Service creates an entry in the state table for each CDC session. These entries can
comprise more than one row. CDC sessions with heterogeneous target tables have state table entries in each
unique relational target database and an entry in a state file on the PowerCenter Integration Service machine for
each nonrelational target. For example, a CDC session that targets Oracle and SQL Server tables and a MQ
Series queue has an entry in the state table in the target Oracle database, in the state table in the target SQL
Server database, and in the state file on the PowerCenter Integration Service machine.
Each session entry in a state table contains a number of repository identifiers and execution state data such as the
checkpoint number and CDC restart information. The following columns can contain PWXPC-specific restart
information:
¨ APPL_ID. Contains the value the PWXPC creates by appending the task instance ID of the CDC session to the
value that you specify in the Application Name attribute in the source PWX CDC application connection. When
this value matches an APPL_ID value for a row in the state table, the PowerCenter Integration Service, in
conjunction with PWXPC, selects the row from the state table for the CDC session.
¨ STATE_DATA. Contains the restart information for the session in a variable-length, 1,024-byte binary column.
When the PowerCenter Integration Service commits change data is to the targets tables, it also commits the
restart information for that data in this column. PWXPC uses the restart information from this column to perform
restart processing for the CDC session.
If the amount of restart information for a session exceeds 1,024 bytes, the PowerCenter Integration Service
adds additional rows to accommodate the remainder of the restart information. For each row added, the
PowerCenter Integration Service increases the value of the SEQ_NUM column by one, starting from zero.
PowerCenter Recovery Files for Nonrelational Targets

If you configure a resume recovery strategy for a CDC session, the PowerCenter Integration Service stores the
session state of operation in the shared location, $PMStorageDir, on the PowerCenter Integration Service
machine. For nonrelational targets, the PowerCenter Integration Service also stores the target recovery status in a

recovery state file in the shared location on the PowerCenter Integration Service machine. PWXPC stores the
restart information for nonrelational target files in this state file.
Recovery State File

For all nonrelational targets in a session, the PowerCenter Integration Service uses a recovery state file on the
PowerCenter Integration Service machine. Nonrelational target files include MQ Series message queues,
PowerExchange nonrelational targets, and other PowerCenter nonrelational targets.
CDC sessions with heterogeneous target tables have state table entries in each unique relational target database
and an entry in a state file on the PowerCenter Integration Service machine for each nonrelational target.
The PowerCenter Integration Service creates the recovery state file in the shared location, $PMStorageDir. The
file name has the following prefix:
pm_rec_state_appl_id
PWXPC creates the value for the appl_id variable in the file name by appending the task instance ID of the CDC
session to the value that you specify in the Application Name attribute in the source PWX CDC application
connection. The PowerCenter Integration Service uses various task and workflow repository attributes to complete
the file name. The message CMN_65003, which the PowerCenter Integration Service writes to the session log,
contains the complete file name.
Application Names
PWXPC, in conjunction with the PowerCenter Integration Service, uses the application name you specify as part of
the key when it stores and retrieves the restart information for the CDC session. When you configure the PWX
CDC application connection for each CDC session, specify a unique value in the Application Name attribute.
PWXPC appends the repository task instance ID for the CDC session to the Application Name value to create the
APPL_ID value in the recovery state table and the appl_id portion in the recovery state file name.
Because the value of the APPL_ID column and the state recovery file contains the task instance ID for the
session, changes to the CDC session such as adding and removing sources or targets affects restart processing.
When you change the CDC session to add or remove sources and targets, you must use the restart token file to
provide restart tokens and then cold start the CDC session.
Restart Processing for CDC Sessions

Each source in a CDC session has its own restart point. The method you use to start a CDC session controls how
PWXPC determines the restart information for the sources in that session.
Use one of the following methods to start CDC sessions:
¨ Cold start. When you cold start a CDC session, PWXPC uses the restart token file to acquire restart tokens for
all sources, does not read the state table or file, and makes no attempt to recover the session. The CDC
session continues to run until stopped or interrupted.
¨ Warm start. When you warm start a CDC session, PWXPC reconciles the restart tokens for sources provided
in the restart token file, if any, with any restart tokens that exist in the state tables or file. If necessary, PWXPC
performs recovery processing. The session continues to run until stopped or interrupted.
¨ Recovery start. When you recover a CDC session, PWXPC reads the restart tokens from any applicable state
tables and file. If necessary, PWXPC performs recovery processing. PWXPC then updates the restart token file
with the restart tokens for each source in the CDC session, and the session ends.
Before you run a CDC session for the first time, you should create and populate the restart token file with restart
tokens for each source in the session. Each restart token pair should match a point in the change stream where
the source and target are in a consistent state. For example, you materialize a target table from a source and do
not change the source data after materialization. To establish a starting extraction, or restart, point in the change

stream, code a special override statement with the CURRENT_RESTART option in the restart token file that has
the file name that you specified in the PWX CDC application connection in the CDC session. When you cold start
the CDC session, PWXPC requests that PowerExchange use the current end-point in the change stream as the
extraction start point. After the CDC session starts, you can resume change activity to the sources.
If you cold start a CDC session and a restart token file does not exist, the PowerCenter Integration Service still
runs the session. Because you did not provide any restart information, PWXPC passes null restart tokens for all
sources to PowerExchange and indicates that the restart tokens for each source are NULL in message
PWXPC_12060. PowerExchange then assigns the default restart point to each source.
Warning: If you use null restart tokens, the CDC session might not produce the correct results. When you cold
start CDC sessions, provide valid restart tokens.
Default Restart Points for Null Restart Tokens

The default restart points that PowerExchange uses when it receives null restart tokens vary by data source type.
The following table describes the default restart points for null restart tokens, by data source type and extraction
method:
Data Source Type Batch and Continuous Extraction Mode Real-time Extraction Mode
All MVS sources Oldest condense file, as recorded in the CDCT. Best available restart point as determined by the
PowerExchange Logger for MVS, which is one of the
following:
- Oldest restart point for which an archive log is
available
- Current active log if there are no available archive
logs.
DB2 for i5/OS Oldest condense file, as recorded in the CDCT. Oldest journal receiver still attached on the journal
receiver chain.
DB2 for Linux, Oldest PowerExchange Logger for Linux, Current log position at the time the PowerExchange
UNIX, and Windows UNIX, and Windows log file, as recorded in the capture catalog was created.
CDCT.
Microsoft SQL Oldest PowerExchange Logger for Linux, Oldest data available in the Publication database.
Server UNIX, and Windows log file, as recorded in the
CDCT.
Oracle Oldest PowerExchange Logger for Linux, Current Oracle catalog dump.
UNIX, and Windows log file, as recorded in the
CDCT.
PowerExchange uses the default restart point only if all sources in a CDC session have null restart tokens. If some
sources have non-null restart tokens, PWXPC assigns the oldest restart point from those tokens to any sources for
which no restart tokens are specified.
For example, a new CDC session contains the sources A, B, and C. The restart token file contains restart tokens
for sources A and B. The restart point for source A is older than that for source B. Source C does not have existing
or supplied restart tokens. Because some sources in the CDC session have explicit restart points, PWXPC does
not assign null restart tokens to source C. Instead, PWXPC assigns the restart point for source A to source C
because this restart point is the oldest one supplied.
Determining the Restart Tokens for Cold Start Processing

When you cold start a CDC session, PWXPC uses the restart token file to determine the restart tokens for all
sources. PWXPC ignores any entries in the state tables or state file for the sources in the CDC session.

More specifically, PWXPC uses one of the following methods to determine the restart tokens:
¨ If the restart token file is empty or does not exist, PWXPC assigns null restart tokens to all sources in the CDC
session.
¨ If the restart token file contains only explicit override statements, PWXPC performs the following processing:
- Assigns the restart tokens in the explicit override statements to the specified sources.
- Assigns the oldest supplied restart point to any sources for which an explicit override statement was not
specified.
¨ If the restart token file contains only the special override statement, PWXPC assigns the restart tokens in the
special override statement to all sources.
¨ If the restart token file contains a special override statement and explicit override statements, PWXPC performs
the following processing:
- Assigns the restart tokens in the special override statement to all remaining sources.
Determining the Restart Tokens for Warm Start Processing

When you warm start a CDC session, uses the state tables and state file, in conjunction with restart token file, to
determine the restart tokens for all sources.
More specifically, PWXPC uses one of the following methods to determine the restart tokens:
¨ If the restart token file is empty or does not exist and there is no matching entry in a state table or state file,
PWXPC assigns null restart tokens to all sources in the session.
¨ If the restart token file is empty or does not exist and if some but not all sources have a matching entry in a
state table or a state file, PWXPC performs the following processing:
- Assigns any restart tokens found in a state table and state file to the appropriate sources.
- Assigns the oldest available restart point to all sources that do not have restart tokens.
¨ If the restart token file is empty or does not exist and if all sources have an entry in a state table or state file,
PWXPC uses the restart tokens from the state tables or state file.
¨ If the restart token file contains explicit override statements and no sources have a matching entry in a state
table or no state file, PWXPC performs the following processing:
- Assigns the oldest supplied restart point to all sources that do not have restart tokens.
¨ If the restart token file contains explicit override statements and if some but not all sources have a matching
entry in a state table or a state file, PWXPC performs the following processing:
- Assigns restart tokens from a state table or state file to the appropriate sources, provided that the tokens
have not been supplied in the restart token file.
- Assigns the oldest available restart point to all sources that do not have restart tokens supplied in the restart
token file or from a state table or state file.
¨ If the restart token file contains explicit override statements and if all sources have an entry in a state table or a
state file, PWXPC performs the following processing:
- Assigns the restart tokens from state tables or the state file to all remaining sources that do not have restart
tokens supplied in the restart token file.

¨ If the restart token file contains only the special override statement, PWXPC assigns the restart tokens in the
special override statement to all sources.
¨ If the restart token file contains a special override statement and explicit override statements, PWXPC performs
the following processing:
- Assigns the restart tokens in the special override statement to all remaining sources.
Group Source Processing in PowerExchange

When you extract change data using PWX CDC application connections, PowerExchange uses group source
processing for all source definitions that you include in a single mapping. With group source processing,
PowerExchange reads data from the same physical source in a single pass. This processing enhances throughput
and reduces resource consumption by eliminating multiple reads of the source data.
When you run a CDC session, PWXPC passes a source interest list that contains all of the sources.
PowerExchange ruses the source interest list to determine the sources for which to read data from the change
stream. When PowerExchange encounters changes for a source in the interest list, it passes the change data to
PWXPC. PWXPC then provides the change data to the appropriate source in the mapping.
If you use PWXPC connections for bulk data movement operations, PowerExchange uses group source
processing for the following multiple-record, nonrelational data sources:
¨ IMS unload data sets
¨ Sequential data sets and flat files
¨ VSAM data sets
PowerExchange uses group source processing to read all records for a single multi-group source qualifier in a
mapping. When you run a bulk data movement session, PWXPC passes PowerExchange the source data map
information from the source definition metadata, which includes the data set or file name if available. If PWXPC
does not pass the data set or file name, PowerExchange determines it from the PowerExchange data map.
PowerExchange reads the data set or file and passes all of the data records to PWXPC. PWXPC then provides
the data records to the appropriate source record type in the multi-group source qualifier.
Using Group Source with Nonrelational Sources

PowerExchange can use group source processing for some nonrelational data sources that support multiple
record types in a single file.
A single mapping can contain one or more multi-record source definitions and single-record source definitions. If
you use PWX NRDB Batch application connections, PWXPC creates a connection to PowerExchange for each
source definition in the mapping and reads the source data.
For data sources with multiple record types, the PowerExchange data map defines a record and a table for each
unique record type. The table represents the relational view of the related record.
For IMS, VSAM, and sequential or flat file data sources, you can use Designer to import data maps with multiple
record types to create PowerCenter source definitions. If you want the source definition to represent only a single
record type, import a single table from the data map. If you want the source definition to include all record types,
import the data map as a multi-record data map.
To import the data map as a multi-record data map, select Multi-Record Datamaps in the Import from
PowerExchange dialog box. If you import a multi-record data map, the source definition has a group for each

table in the data map. A group contains metadata for the fields in the table. If you import a single table from a multi-
record data map, the source definition has only a single group.
When you run a session that contains a mapping with source definitions for each table in a multi-record data map,
PowerExchange reads the data set or file once for each source definition. When you run a session that contains a
mapping with a single source definition for all records in a multi-record data map, PowerExchange uses group
source processing to read all of the records in the data set or file in a single pass.
For example, if you have a sequential file that contains three different record types, you can create a source
definition for each record type. Then create a mapping that contains the three source definitions. When you run a
session that contains the mapping, PowerExchange reads the sequential file three times.
Alternatively, if you import the data map as a multi-record data map and create a single multi-record source
definition, you can use this multi-record source definition in a mapping. When you run a session that contains this
mapping, PowerExchange reads the sequential file one time to extract the data.
When you import IMS data maps as multi-record data maps, you can use the source definitions only to process
IMS unload data sets. You cannot use multi-record IMS source definitions to read all segments from an IMS
database in a single pass. To perform bulk data movement operations on IMS databases, create mappings that
have a source definition for each segment in the IMS database.
Using Group Source with CDC Sources

When you use PWX CDC application connections to extract change data, PowerExchange automatically uses
group source processing and reads the change stream in a single pass for all source definitions in the mapping.
All sources in the mapping must be the same data source type and must read the same change stream.
To create source definitions in Designer that can be used to extract change data, import source metadata by using
one of the following methods:
¨ Import a PowerExchange extraction map by using the Import from PowerExchange dialog box.
¨ Import the table definitions from relational databases, by using either the Import from PowerExchange dialog
box or the Import from Database dialog box.
Restriction: To read change data for nonrelational sources, you must import extraction maps from
PowerExchange.
Informatica recommends that you use extraction maps to create source definitions for all CDC sources. When you
create source definitions from extraction maps, the mapping and session creation process is simpler for the
following reasons:
¨ The source definition contains the extraction map name, which eliminates the need to provide it when you
configure the session.
¨ The source definition contains the PowerExchange-defined CDC columns, which eliminates the need to add
these columns to the source definition. The PowerExchange-defined columns include the change indicator and
before image columns as well as the DTL__CAPX columns.
When you extract change data, PowerExchange uses group source processing for all source definitions in the
mapping. All source definitions must be for the same data source type, such as DB2, IMS, VSAM, or Oracle. Do
not include multiple data source types in the mapping. Otherwise, the session fails with message PWXPC_10080.
For example, you cannot run a CDC session that contains a mapping with both VSAM and IMS source definitions,
even though the change stream is the same. To extract change data for both IMS and VSAM data sources, create
unique a mapping and session for the VSAM sources and a separate, unique mapping and session for the IMS
sources. PowerExchange reads the change stream twice, once for the session with VSAM sources and once for
the session with IMS sources.
Group Source Processing in PowerExchange 117

If you create a workflow that contains multiple CDC sessions, PowerExchange uses a connection for each
session, even if the sessions extract change data from the same change stream, such as the PowerExchange
Logger for MVS.
The following example mapping shows three DB2 sources, for which the source definitions were created from
extraction maps:
If you include this mapping in a session that uses a PWX DB2zOS CDC application connection, PowerExchange
uses group source processing to read the change stream and extract the changes for all three source tables.
PowerExchange extracts change data in chronological order, based on when the UOWs were completed.
PowerExchange passes the change data to PWXPC, and PWXPC provides the changes to the appropriate source
qualifier.
Note: Because the example mapping uses source definitions created from extraction maps, it cannot be used for
bulk data movement operations. However, mappings that use source definitions created from database relational
metadata can be used for either change data extraction or bulk data movement.
Commit Processing with PWXPC

The PowerCenter Integration Service, in conjunction with PWXPC, commits data to the target based on commit
properties and the commit type. Commit properties specify the commit interval and the number of UOWs or
change records that you want to use as a basis for the commit. The commit type determines when the
PowerCenter Integration Service commits data to the target.
By default, the Commit Type attribute on the session Properties tab specifies Target, which indicates target-
based commit processing. For CDC sessions, the PowerCenter Integration Service always uses source-based
commit processing, and PWXPC controls the timing of commit processing. When you run a CDC session that
specifies target-based commit processing, the PowerCenter Integration Service automatically changes the commit
type to source-based and writes message WRT_8226 in the session log.
PWXPC ignores the Commit Interval attribute. To control commit processing, configure attributes on the PWX
CDC Change and Real Time application connections.

RELATED TOPICS:
¨ “Commitment Control Options” on page 133
Controlling Commit Processing

To control commit processing, you can specify certain PWX CDC Real Time or Change application connection
attributes.
The following table describes the connection attributes that control commit processing:
Connection Attribute Real Time or Description

Change
Connections
Maximum Rows Per commit Both Maximum number of change records that PWXPC processes before
it flushes the data buffer to commit the change data to the targets. If
necessary, PWXPC continues to process change records across
UOW boundaries until this maximum rows threshold is met. PWXPC
does not wait for a UOW boundary to commit the change data.
Default is 0, which means that PWXPC does not use maximum rows.
Minimum Rows Per commit Real Time Minimum number of change records that PowerExchange reads from
the change stream before it passes any commit records in the
change stream to PWXPC. Before reaching this minimum value,
PowerExchange skips commit records and passes only the change
records to PWXPC.
Default is 0, which means that PowerExchange does not use
minimum rows.
Real-time Flush Latency in milli- Real Time Number of milliseconds that must pass before PWXPC flushes the
seconds data buffer to commit the change data to the targets. When this
latency period expires, PWXPC continues to read the changes in the
current UOW until the end of that UOW is reached. Then, PWXPC
flushes the data buffer to commit the change data to the targets.
Default is 0, which means that PWXPC uses 2,000 milliseconds.
UOW Count Both Number of UOWs that PWXPC processes before it flushes the data
buffer to commit the change data to the targets.
Default is 1.
You can specify values for the all of these commitment control attributes. However, PWXPC commits change data
only when one of the following values is met:
¨ Maximum Rows Per commit

¨ Real-time Flush Latency in milli-seconds
¨ UOW Count
If you specify a value for the Minimum Rows Per commit attribute, this threshold must be met before a commit
can occur. However, PWXPC flushes the data buffer to commit the change data to the targets only when
Maximum Rows Per commit, Real-time Flush Latency in milli-seconds, or UOW Count is met, whichever is
first.
After PWXPC commits the change data, it resets the UOW count, the maximum and minimum rows, and the real-
time flush latency timer. PWXPC continues to read change data. Whenever one of the commitment control values
is met, PWXPC commits that data to the targets. Commit processing continues until the CDC session is stopped,
ends, or terminates abnormally. When the PWXPC CDC reader ends normally, PWXPC issues a final commit to
Commit Processing with PWXPC 119

flush all complete, buffered UOWs and their final restart tokens to the targets. Prior to ending, the PWXPC CDC
reader writes the following message to the session log:
PWXPC_12075 [INFO] [CDCRestart] Session complete. Next session will restart at: Restart 1
[restart1_token] : Restart 2 [restart2_token]
Restriction: If you enable the Commit On End Of File attribute on the session Properties tab, duplicate data can
occur in the targets because the PowerCenter Integration Service commits any remaining change data in the
buffer to the targets. This final commit by the PowerCenter Integration Service occurs after the PWXPC CDC
reader has committed all complete UOWs in the buffer, along with their restart tokens, to the targets. As a result,
the final restart tokens might represent a point in the change stream that is earlier than final change data that the
PowerCenter Integration Service commits to the targets. To prevent possible duplicate data when you restart CDC
sessions, set the Commit Type attribute to Source and disable the Commit On End Of File attribute.
Maximum and Minimum Rows per Commit

The Maximum Rows Per commit attribute controls the size of the UOWs written to the targets. The
Minimum Rows Per commit attribute controls the size of the UOWs read from the change stream. You can use
these attributes to mitigate the effects of processing very small or very large UOWs.
Maximum Rows per Commit

If you have very large UOWs, you can use the Maximum Rows Per commit attribute to specify the maximum
number of change records that PWXPC reads before it commits the change data to the targets. This attribute
causes PWXPC to commit change data without waiting for a UOW boundary, which is called a subpacket commit.
By using a subpacket commit for large UOWs, you can minimize storage use on the PowerCenter Integration
Service machine and lock contention on the target databases.
Warning: Because PWXPC can commit change data to the targets between UOW boundaries, relational integrity
(RI) might be compromised. Do not use this connection attribute if you have targets in the CDC session with RI
constraints.
Generally, you should use the maximum rows attribute only if you have large UOWs that cannot be processed
without impacting either the PowerCenter Integration Service machine or the target databases. For example, if you
have an application that makes 100,000 changes before it issues a commit, you can use the maximum rows
attribute to commit the change data before PWXPC reads all 100,000 change records. When the maximum rows
limit is met, PWXPC flushes the change data from the buffer on the PowerCenter Integration Service machine and
commits the data to the targets. After the commit processing, the RDBMS can release the locks in the target
databases for these change records and the PowerCenter Integration Service can reuse the buffer space for new
change records.
Minimum Rows per Commit

If your change data has many small UOWs, you can use the Minimum Rows Per commit attribute to create
larger UOWs of a more uniform size. Use this attribute to specify the minimum number of change records that
PowerExchange must pass to PWXPC before passing a commit record. Until the minimum rows value is met,
PowerExchange discards any commit records that it reads from the change stream and passes only change
records to PWXPC. After the minimum rows limit is met, PowerExchange passes the next commit record to
PWXPC and then resets the minimum rows counter.
Online transactions that run in transaction control systems such as CICS and IMS often commit after making only
a few changes, which results in many, small UOWs in the change stream. PowerExchange and PWXPC can
process fewer, larger UOWs more efficiently than many small UOWs. Therefore, if you use the minimum rows limit
to increase the size of UOWs, you can improve CDC processing efficiency.
A minimum rows limit does not impact the relational integrity of the change data because PowerExchange does
not create new commits points in the change stream data. PowerExchange simply skips some of the original
commit records in the change stream.

Target Latency
Target latency is the total time that PWXPC uses to extract change data from the change stream and that the
PowerCenter Integration Service uses to apply that data to the targets. If this processing occurs quickly, target
latency is low.
The values you select for the commitment control attributes affect target latency. You must balance target latency
requirements with resource consumption on the PowerCenter Integration Service machine and the target
databases.
Lower target latency results in higher resource consumption because the PowerCenter Integration Service must
flush the change data more frequently and the target databases must process more commit requests.
You can affect target latency by setting the commit control attributes.
The following default values can result in the lowest latency:
¨ 0 for Maximum Rows Per commit, which disables this option

¨ 0 for Minimum Rows Per commit, which disables this option
¨ 0 for Real-time Flush Latency in milli-seconds, which is equivalent to 2000 milliseconds or 2 seconds
¨ 1 for UOW Count
These values can decrease target latency because PWXPC commits changes after each UOW, or on UOW
boundaries. However, these values also cause the highest resource consumption on the source system, the
PowerCenter Integration Service machine, and the target databases. Alternatively, these values might decrease
throughput because change data flushes too frequently for the PowerCenter Integration Service or the target
databases to handle.
To lower resource consumption and potentially increase throughput for CDC sessions, specify a value greater than
the default value for only one of the following attributes:
¨ Maximum Rows Per commit
¨ UOW Count
¨ Real-time Flush Latency in milli-seconds
Disable the unused attributes.
Examples of Commit Processing

The following examples show how the commitment control attributes affect commit processing with PWXPC.
Subpacket Commit and UOW Count - Example

This example uses the Maximum Rows Per commit and UOW Count attributes to control commit processing.
The change data is composed of UOWs of the same size. Each UOW contains 1,000 change records. The
commitment control attributes have the following values:
¨ 300 for Maximum Rows Per commit
¨ 0 for Minimum Rows Per commit, which disables this attribute
¨ 0 for Real-time Flush Latency in milli-seconds, which is equivalent to 2 seconds
¨ 1 for UOW Count
Based on the maximum rows value, PWXPC flushes the data buffer after reading the first 300 records in a UOW.
This action commits the change data to the targets. PWXPC continues to commit change data to the targets every
300 records.
Commit Processing with PWXPC 121

PWXPC commits on UOW boundaries only for the UOW count and real-time flush latency interval. If the real-time
flush latency interval expires before PWXPC reads 300 change records, PWXPC still commits based on the
maximum rows value because that threshold is met before a UOW boundary occurs.
When the end of the UOW is read, PWXPC commits the change data because the UOW Count value is 1.
PWXPC resets the UOW and maximum row counters and the real-time flush latency timer each time it commits.
Because all of the UOWs have the same number of change records, PWXPC continues to read change data and
to commit the data to the targets at the same points in each UOW.
In this example, PWXPC commits change data at the following points:
¨ 300 change records based on the maximum rows value

¨ 1,000 change records based on the UOW count value
UOW Count and Time-Based Commits - Example

This example uses the UOW Count and Real-time Flush Latency in milli-seconds attributes to control commit
processing. The change data consists of UOWs of varying sizes. The commitment control attributes have the
following values:
¨ 0 for Maximum Rows Per commit, which disables this attribute
¨ 0 for Minimum Rows Per commit, which disables this attribute
¨ 5000 for Real-time Flush Latency in milli-seconds, which is equivalent to 5 seconds
¨ 1000 for UOW Count
Initially, PWXPC reads 900 complete UOWs in 5 seconds. Because the real-time flush latency interval has
expired, PWXPC flushes the data buffer to commit the change data to the targets. PWXPC then resets both the
UOW counter and real-time flush latency timer. When PWXPC reaches UOW 1,000, PWXPC does not commit
change data to the targets because the UOW counter was reset to 0 after the last commit.
PWXPC reads the next 1,000 UOWs in 4 seconds, which is less than the real-time flush latency timer. PWXPC
commits this change data to the target because the UOW counter has been met. After this commit, PWXPC then
resets the real-time flush latency timer and the UOW counter.
PWXPC continues to read change data and commit the data to the targets, based on the UOW count or the real-
time flush latency flush time, whichever limit is met first.
In this example, PWXPC commits change data at the following points:
¨ After UOW 900 because the real-time latency flush latency timer matched first
¨ After UOW 1,900 because the UOW count matched first during the second commit cycle
Minimum Rows and UOW Count - Example

This example uses the Minimum Rows Per commit and UOW Count attributes to control commit processing. The
change data consists of UOWs of the same size. Each UOW contains ten change records. The commitment
control attributes have the following values:
¨ 0 for Maximum Rows Per commit, which disables this attribute
¨ 100 for Minimum Rows Per commit
¨ -1 for Real-time Flush Latency in milli-seconds, which is disables this attribute
¨ 10 for UOW Count

PWXPC passes the minimum rows value to PowerExchange and requests change data from the change stream.
Because the minimum rows value is 100, PowerExchange skips the commit records of the first nine UOWs. When
PowerExchange reads the last change record in the tenth UOW, the minimum rows limit is met. So,
PowerExchange passes the commit record for the tenth UOW to PWXPC and resets the minimum rows counter.
PWXPC increases the UOW counter to one.
PowerExchange and PWXPC continue to read the change data until the UOW counter is 10. At this point, PWXPC
flushes the data buffer to commit the change data to the targets and resets the UOW counter.
In this example, PWXPC commits change data after 1,000 change records, which is also after every 10 UOWs
because each UOW contains 10 change records and the UOW Count is 10.
Offload Processing
You can use CDC offload processing and multithreaded processing to improve performance and efficiency of real-
time CDC sessions.
You can use CDC offload processing to distribute processing to the PowerCenter Integration Service machine
running the extraction, which reduces processing on the source system. You can also use CDC offload processing
to copy change data to a remote system by using the PowerExchange Logger for LINUX, UNIX, and Windows.
You can use multithreaded processing to increase parallelism on the PowerCenter Integration Service machines.
CDC Offload Processing

When you extract change data, PowerExchange maps the captured data to the columns in the extraction map.
PowerExchange also performs any data manipulation operations that you defined in the extraction map, such as
populating change-indicator and before-image columns or running expressions. This column-level processing of
change data occurs in the PowerExchange Listener and can be CPU-intensive.
By default, PowerExchange performs column-level processing on the system on which the changes are captured.
For MVS, DB2 for i5/OS, and Oracle sources, PowerExchange also runs the UOW Cleanser to reconstruct
complete UOWs from the change data in the change stream on the system.
To reduce the overhead of column-level and UOW Cleanser processing, you can use CDC offload processing.
CDC offload processing moves the column-level and UOW Cleanser processing to the PowerCenter Integration
Service machine running the extraction. CDC offload processing can also be used by the PowerExchange Logger
for Linux, UNIX, and Windows to copy change data to PowerExchange Logger log files on a remote system. You
can then extract the change data from the remote system rather than the original source system.
Use CDC offload processing to help increase concurrency and throughput and decrease costs in the following
situations:
¨ You have insufficient resources on the machine where the change data resides to run the number of concurrent
extraction sessions you require.
¨ You have insufficient resources on the machine where the change data resides to provide the necessary
throughput you require.
¨ You have spare cycles on the PowerCenter Integration Service machine and those cycles are cheaper than the
cycles on the machine on which the changes are captured.
Offload Processing 123

Multithreaded Processing
If you use CDC offload processing for change data extractions, you can also use multithreaded processing, which
might improve help improve throughput even more. By default, PowerExchange performs column-level processing
on the change stream as a single thread. If you use multithreaded processing, PowerExchange might be able to
extract changes faster and more efficiently by processing more than one UOW simultaneously.
PowerExchange multithreaded processing splits a UOW into multiple threads on the PowerCenter Integration
Service machine. After the column-level processing completes, PowerExchange merges the threads and passes
the UOW to the PWXPC CDC reader for processing. Multithreaded processing works most efficiently when
PowerExchange on the source machine is supplying data fast enough to take full advantage of the multiple
threads on the PowerCenter Integration Service machine. If PowerExchange completely utilizes a single processor
on the PowerCenter Integration Service machine, then multithreaded processing may provide increased
throughput.

CHAPTER 8
Extracting Change Data

¨ Overview of Extracting Change Data, 125

¨ Task Flow for Extracting Change Data, 126
¨ Testing a Change Data Extraction, 126
¨ Configuring PowerCenter CDC Sessions, 128
¨ Creating Restart Tokens for Extractions, 135
¨ Displaying Restart Tokens, 135
¨ Configuring the Restart Token File, 136
Overview of Extracting Change Data

Use PowerExchange in conjunction with PWXPC and PowerCenter to extract captured change data and write the
data to one or more targets. To extract change data that PowerExchange captures, you must import metadata for
the CDC sources and the targets of the change data in Designer. After creating the source and target definitions in
Designer, you must create a mapping and then an application connection, session, and workflow in Workflow
Manager. You can create multiple mappings, sessions, and workflows based on the same source and target
definitions, if appropriate.
For relational data sources, you can import the metadata from either database definitions or PowerExchange
extraction maps. For nonrelational sources, you must import PowerExchange extraction maps.
Tip: Informatica recommends that you import the metadata from PowerExchange extraction maps instead of from
database definitions. When you import extraction maps, the source definition contains all of the PowerExchange-
generated CDC columns, such as the before image (BI) and change indicator (CI) columns. Additionally, PWXPC
derives the extraction map name from the source definition so you do not need to code the extraction map name
for each source in the session properties.
Before starting a CDC session, you should create restart tokens to define an extraction start point in the change
stream. Restart tokens might also be required for resuming extraction processing in a recovery scenario.
To stop a CDC session using real-time extraction mode based on certain user-defined events, you can configure
event table processing. Also, you can offload column-level extraction processing and any UOW Cleanser
processing from the source system to the following remote locations:
¨ PowerCenter Integration Service machine
¨ A remote machine where the PowerExchange Logger for Linux, UNIX, and Windows runs
If you use offload processing with real-time extractions, you can also use multithreaded processing.
125
Task Flow for Extracting Change Data
Perform the following tasks in the PowerExchange Navigator, PowerCenter Designer, and PowerCenter Workflow
Manager to configure and start extraction processing.
Before you begin, complete configuration of the data source and PowerExchange for CDC, and create capture
registrations in the PowerExchange Navigator.
1. Edit the extraction map if necessary.

You can make the following changes:
¨ Deselect any column for which you do not want to extract the change data. PowerExchange still captures
change data for these columns.
¨ Add change indicator (CI) and before image (BI) columns.
2. To test the extraction map, perform a database row test on the extraction map in PowerExchange Navigator.
3. In Designer, import metadata for the sources and targets.
4. In Designer, configure a mapping to extract and process change data.
5. In Workflow Manager, configure a connection and session.
6. Create restart tokens for the CDC session.
7. Configure the restart token file.
8. If you want to stop extraction processing based on certain events, implement event table processing.
9. If you want to offload column-level extraction processing and UOW Cleanser processing from the source
system to the PowerCenter Integration Service machine or PowerExchange Logger for Linux, UNIX, and
Windows machine, configure offload processing. For real-time extractions, you can also configure
multithreaded processing.
10. Start the CDC session.
RELATED TOPICS:
¨ “Starting PowerCenter CDC Sessions” on page 140
¨ “Monitoring and Tuning Options” on page 148
¨ “Testing a Change Data Extraction” on page 126
Testing a Change Data Extraction

Perform a database row test in the PowerExchange Navigator to ensure that PowerExchange can retrieve data
when the extraction map is used in a CDC session.
A database row test verifies that:
¨ PowerExchange has captured change data for a data source defined in a capture registration.
¨ PowerExchange Condense or the PowerExchange Logger for Linux, UNIX, and Windows has captured change
data for a capture registration, if applicable.
¨ The extraction map properly maps the captured change data.
126 Chapter 8: Extracting Change Data

To test change data extraction:
1. In the Resource Explorer of the PowerExchange Navigator, open the extraction group that includes the
extraction map that you want to test.
2. Open the extraction map.
3. Select the extraction map and click File > Database Row Test.
4. In the Database Row Test dialog box, enter or edit the following information:
Field Description
DB Type An extraction mode indicator:

- CAPXRT. Real-time extraction mode or continuous extraction mode.
- CAPX. Batch extraction mode.
Location Node name for the location of the system on which the captured change data resides. This
name must be defined in a NODE statement in the dbmover.cfg file on the Windows machine
from which you run the database row test.
UserID and Password Optionally, a user ID and password that provides access to the source change data.
Application Name At least one character to represent the application name. For a row test, a unique application
name is not required. PowerExchange does not retain the value that you specify.
SQL Statement A SQL SELECT statement that PowerExchange generates for the fields in the extraction map.
You can edit this statement, if necessary.
In the statement, a table is identified in the following format:
Schema.RegName_TableName
Where:
- Schema is schema for the extraction map.
- RegName is the name of the capture registration that corresponds to the extraction map.
- TableName is the table name of the data source.
Note: If you enter CAPX in the DB Type field, you can only extract change data after PowerExchange
Condense or the PowerExchange Logger for Linux, UNIX, and Windows has closed at least one condense or
log file. Otherwise, PowerExchange displays no data in PowerExchange Navigator and writes the PWX-04520
message in the PowerExchange message log on the extraction system. PowerExchange also writes this
message if no change data for the data source has been captured, condensed, or logged.
5. Click Advanced.
6. In the CAPX Advanced Parameters or CAPXRT Advanced Parameters dialog box, enter information,
including the following:
¨ If you use continuous extraction mode, enter the CAPX CAPI_CONNECTION name in the CAPI
Connection Name field.
¨ If you use the PowerExchange Logger for Linux, UNIX, and Windows to offload change data to system
remote from the system on which it was captured, enter location of the extraction maps in the Location
field.
7. Click OK.
8. Click Go.
The database row test returns each change from the extraction start point by column. The results include the
PowerExchange-defined CDC columns, the DTL__ columns, which provide information such as the change
type, change timestamp, and user ID of the user who made the change.
Testing a Change Data Extraction 127

Configuring PowerCenter CDC Sessions
After you import metadata for CDC data sources and targets into PowerCenter, you can create a mapping and a
CDC session to extract change data. Before running CDC sessions, you must configure numerous session and
connection attributes.
Changing Default Values for Session and Connection Attributes

Certain PowerCenter session and application connection attributes have default values that are only appropriate
for bulk data movement. You must change the values of these attributes for CDC sessions.
The following table summarizes these attributes and their recommended values:
Attribute Name Attribute Recommended Description

Location Value
Commit Type Properties Tab Source Default is Target. The PowerCenter Integration Service
automatically overrides it to Source. However, you cannot
disable Commit On End Of File unless you change Commit
Type to Source.
Commit On End Properties Tab Disabled Default is enabled. The PowerCenter Integration Service
Of File performs a commit when the session ends. This commit
occurs after PWXPC commits the restart tokens, which can
cause an out-of-sync condition between the restart tokens
and the target data. As a result, duplicate data can occur
when CDC sessions restart.
Recovery Properties Tab Resume from last Default value is Fail task and continue workflow. To properly
Strategy checkpoint restart CDC session, PowerExchange CDC and PWXPC
require that this option is set to Resume from last
checkpoint.
Stop on errors Config Object Tab 1 Default value is 0. By default, the PowerCenter Integration
Service does not consider errors when writing to targets as
fatal. The following types of error are non-fatal:
- Key constraint violations
- Loading nulls into a not null field
- Database trigger responses
If write errors occur, you might experience change data loss
because PWXPC has advanced the restart tokens values.
To maintain target data and restart token integrity, you must
set this option to 1.
Application Name Application Code a unique Default is the first 20 characters of the WorkFlow Name.
Connection name for each CDC Warning: The default might not result in a unique name.
session.
RestartToken Application Default value Use the default value of $PMRootDir/Restart, which PWXPC
File Folder Connection creates if it does not exist.

Attribute Name Attribute Recommended Description
Location Value
RestartToken Application Code a unique If no value is entered for Application Name, the default is
File Name Connection name for each CDC the workflow name. Otherwise, the value for Application
session. Name is used.
Warning: The default may not result in a unique name.
Number of Runs Application 1 or higher Default is 0. PWXPC keeps only one backup copy of the
to Keep Connection restart token initialization and termination files. Specify a
RestartToken File value greater than 0 so a history is available for recovery
purposes.
Configuring Application Connection Attributes

To extract change data, you must configure certain application connection attributes. For a complete list of all
PWX CDC application connection attributes, see PowerExchange Interfaces for PowerCenter.
Image Type
For update operations, use the Image Type attribute to configure the format of the change data that a CDC
session extracts.
Select one of the following options for the Image Type attribute:
¨ AI. After images only.
¨ BA. Before and after images.
Default is BA.
If you select BA for the Image Type attribute, PowerExchange provides the before-image (BI) and after-image (AI)
data for the updated row as separate SQL operations:
¨ A DELETE with the before-image data
¨ An INSERT with the after-image data
Note: To select BA with batch or continuous extraction mode, you must configure PowerExchange Condense or
the PowerExchange Logger for Linux, UNIX, and Windows to log before and after images. Otherwise, you can only
select after images.
If you select AI for the Image Type attribute, PowerExchange provides the after-image data for updated row as a
SQL UPDATE operation.
You can also configure one or more data columns in an extraction map with before-image (BI) columns. Use the
PowerExchange Navigator to update the extraction map with before-image columns, which adds additional
columns to the extraction map with the name of DTL__BI_columnname. If you use BI columns, select AI for the
Image Type attribute. PowerExchange then includes before-image data in any BI columns, along with the after-
image data, in a single SQL UPDATE operation.
When you configure BI columns, you can make decisions about UPDATE operations in a mapping because the
before and after-image data is contained in a single record. For example, you can use BI columns to handle
update operations that change the value of a key column of a row. Some relational databases, such as DB2 for z/
OS, allow update operations to key columns. The RDBMS understands that this operation is equivalent to deleting
the row and then re-adding it with a new primary key and logs the change as an update.
If you select AI for the Image Type attribute, PowerExchange provides these changes as an UPDATE operation.
Because some relational databases do not allow updates to primary key columns, you cannot apply these changes
Configuring PowerCenter CDC Sessions 129

as updates. If you configure BI columns for key columns, you can then use the Flexible Key Custom
transformation to be change any UPDATE operations for key columns into a DELETE operation followed by an
INSERT operation.
Event Table Processing

You can use event table processing to stop the extraction of changes based on user-defined events, such as an
end-of-day event. For example, to stop an extraction process every night, after all of the changes for the day have
been processed, write a change to the event table at midnight. This change triggers PowerExchange to stop
reading change data and shut down the extraction process after the current UOW completes.
Event table processing has the following rules and guidelines:
¨ You can only use event table processing with real-time or continuous extraction modes.
¨ You must create the event table, and define the applications that can update the table.
¨ You must register the event table for change data capture from the PowerExchange Navigator.
¨ A CDC session monitors a single event table. Each user-defined event requires its own event table and a
separate extraction process.
¨ The event table and all of the source tables in the CDC session must be of the same source type.
To implement event table processing:
1. Create an event table.

The event table must be of the same source type and on the same machine as the change data that is
extracted. For example, if you extract DB2 change data on MVS, the event table must be a DB2 table in the
same DB2 subsystem as the DB2 source tables for the extraction.
2. In the PowerExchange Navigator, create a capture registration and extraction map for the event table.
When you create a capture registration, the PowerExchange Navigator generates an extraction map.
3. In PowerCenter, create a CDC session, and specify the extraction map name in the Event Table attribute on
the PWX CDC Real Time application connection.
4. When the defined event occurs, update the event table.
When PowerExchange reads the update to the event table, PowerExchange places an end-of-file (EOF) into
the change stream. PWXPC processes the EOF, passes it to the PowerCenter Integration Service, and then
shuts down the PowerExchange reader. The PowerCenter Integration Service completes writing all of the
data currently in the pipeline to the targets and then ends the CDC session.
CAPI Connection Name Override

PowerExchange allows a maximum of eight CAPI_CONNECTION statements in the DBMOVER configuration file.
You can use multiple CAPI_CONNECTION statements to extract changes from more than one data source type
with a single PowerExchange Listener on a single machine. For example, you can extract changes for Oracle and
DB2 for Linux, UNIX, and Windows through a single PowerExchange Listener by specifying multiple
CAPI_CONNECTION statements in the dbmover.cfg file.
To specify the CAPI_CONNECTION statement that PowerExchange uses to extract change data in a CDC
session, code the name in the CAPI Connection Name Override attribute.
You must code CAPI_CONNECTION statements on the system where the change data resides so that
PowerExchange can extract change data for a data source type. If you use CDC offload processing, you must also
code the CAPI_CONNECTION statements in the dbmover.cfg file on the PowerCenter Integration Service machine.

Idle Time
To indicate whether a real-time or continuous extraction mode CDC session should run continuously or shutdown
after reaching the end-of-log (EOL), use the Idle Time attribute.
Enter one of the following values for the Idle Time attribute:
¨ -1. The CDC session runs continuously. PowerExchange returns end-of-file (EOF) only when the CDC session
is manually stopped.
¨ 0. After reaching EOL, PowerExchange returns EOF and the CDC session ends.
¨ n. After reaching EOL, PowerExchange waits for n seconds and, if no new change data of interest arrives, the
CDC session ends. Otherwise, the CDC session continues until PowerExchange waits for n seconds without
reading new change data of interest.
Default is -1.
PowerExchange determines the EOL by using the current end of the change stream at the point that
PowerExchange started to read the change stream. PowerExchange uses the concept of EOL because the
change stream is generally not static, and so the actual end-of-log is continually moving forward. After
PowerExchange reaches EOL, it writes the PWX-09967 message in the PowerExchange message log.
Typically, real-time and continuous extraction mode CDC sessions use the default value of -1 for the Idle Time
attribute. If necessary, you can manually stop a never-ending CDC session by using the PowerCenter Workflow
Monitor, pmcmd commands, or the PowerExchange STOPTASK command.
Alternatively, you can set the Idle Time attribute to 0. After PowerExchange reaches EOL, it returns an EOF to
PWXPC. PWXPC and the PowerCenter Integration Service then perform the following processing:
1. PWXPC flushes all buffered UOWs and the ending restart tokens to the targets.
2. The CDC reader ends.
3. After the PowerCenter Integration Service finishes writing the flushed data to the targets, the writer ends.
4. After any post-session commands and tasks execute, the CDC session ends.
If you set the Idle Time attribute to a positive number, the following processing occurs:
1. PowerExchange reads the change stream until it reaches EOL, and then timing for the idle time begins.
2. If more data is in the change stream after EOL, PowerExchange continues to read the change stream, looking
for change data of interest to the CDC session, as follows:
¨ If the idle time expires before PowerExchange reads a change record of interest for the CDC session,
PowerExchange stops reading the change stream.
¨ If PowerExchange reads a change record of interest to the CDC session, PowerExchange restarts the
timer, passes the change data to PWXPC, and continues to read the change stream. This processing
continues until the idle time expires.
3. After the idle time expires, PowerExchange passes an EOF to PWXPC.
4. PWXPC and the PowerCenter Integration Service perform the same processing as when the Idle Time
attribute is set to 0 and the CDC session ends.
If you set the Idle Time attribute to a low value, the CDC session might end before all available change data in the
change stream has been read. If you want a CDC session to end periodically, Informatica recommends that you
set the Idle Time attribute to 0 because active systems are rarely idle.
When a CDC session ends because either the idle time value has been reached or a PowerExchange STOPTASK
command has been issued, PWXPC writes the following message in the session log:
[PWXPC_10072] [INFO] [CDCDispatcher] session ended after waiting for [idle_time] seconds. Idle Time
limit is reached
If you stop a never-ending CDC session with the PowerExchange STOPTASK command, PWXPC substitutes
86400 for the idle_time variable in the PWXPC_10072 message.

Note: If you specify values for the Reader Time Limit and Idle Time attributes, the PowerCenter Integration
Service stops reading data from the source when the first one of these terminating conditions is reached. Because
the reader time limit does not result in normal termination of a CDC session, Informatica recommends that you use
only the idle time limit.
Restart Control Options

PWXPC uses the restart information to tell PowerExchange from which point to start reading the captured change
data. To specify restart information, PWXPC provides options that you must configure for each CDC session.
The following table describes the restart attributes you must configure for CDC sessions:
Connection Attribute Description
Application Name Application name for the CDC session. Specify a unique name for each CDC session. The
application name is case sensitive and cannot exceed 20 characters.
Default is the first 20 characters of the workflow name.
RestartToken File Folder Directory name on the PowerCenter Integration Service machine that contains the restart
token override file.
Default is $PMRootDir/Restart.
RestartToken File Name File name in the RestartToken File Folder that contains the restart token override file.
PWXPC uses the contents of this file, if any, in conjunction with the state information to
determine the restart point for the CDC session.
Default is the Application Name, if specified, or the workflow name, if Application Name is not
specified.
Informatica recommends that you specify a value for the Application Name attribute, because the default value
might not result in a unique name. The values for Application Name and RestartToken File Name attributes
must be unique for every CDC session. Non-unique values for either of these attributes can cause unpredictable
results that include session failures and potential data loss.
PowerExchange Flush Latency

PowerExchange reads change data into a buffer on the source machine, or on the PowerCenter Integration
Service machine if you use CDC offload processing. The PowerExchange Consumer API (CAPI) interface flushes
the buffer that contains the data to PWXPC on the PowerCenter Integration Service machine for processing when
the one of the following conditions occurs:
¨ The buffer becomes full.
¨ The CAPI interface timeout, also called the PowerExchange flush latency, expires.
¨ A commit point occurs.
PowerExchange uses the flush latency value as the CAPI interface timeout value on the source machine, or on the
PowerCenter Integration Service machine if you use CDC offload processing.
For CDC sessions that use real-time or continuous extraction mode, set the flush latency in the
PWX Latency in seconds attribute of the PWX CDC Real Time application connection. For CDC sessions that
use batch extraction mode, PowerExchange always uses two seconds for the flush latency.
Restriction: The value of PWX Latency in seconds impacts the speed with which a CDC session responds to a
stop command from Workflow Monitor or pmcmd, because PWXPC must wait for PowerExchange to return
control before it can handle the stop request. Informatica recommends that you use the default value of 2 seconds
for the PWX Latency in seconds attribute.

PowerExchange writes the message PWX-09957 in the PowerExchange message log to reflect the CAPI interface
timeout value set from the flush latency value. If you select Retrieve PWX Log Entries on the application
connection, PWXPC also writes this message in the session log.
After PowerExchange flushes the change data to PWXPC, PWXPC provides the data to the appropriate sources in
the CDC session for further processing and the PowerCenter Integration Service commits the data to the targets.
Commitment Control Options

PWXPC, in conjunction with PowerExchange and the PowerCenter Integration Service, controls the timing of
commit processing for CDC sessions based on the values you code for the commitment control options.
To control commit processing, set one or more of the following connection attributes:
Maximum Rows Per commit
Maximum number of change records in a source UOW that PWXPC processes before it flushes the data
buffer to commit the change data to the targets. If necessary, PWXPC continues to process change records
across UOW boundaries until the maximum rows limit is met. PWXPC does not wait for a UOW boundary to
commit the change data. After the maximum rows limit is met, PWXPC issues a real-time flush to commit the
change data and the restart tokens to the targets and writes the PWXPC_12128 message to the session log.
PWXPC resets the maximum rows limit when a real-time flush occurs because either the maximum rows limit
or UOW count is met or the real-time flush latency timer expires.
Note: The Maximum Rows Per commit attribute is a count of records within a UOW, unlike the UOW Count
attribute that is a count of complete UOWs.
Default is 0, which means that PWXPC does not use maximum rows.
PWXPC uses the maximum rows limit to commit data before an end-UOW is received, a process also called
sub-packet commit. If you specify either 0 or no value, commits occur only on UOW boundaries. Otherwise,
PWXPC uses the value that you specify to commit change records between UOW boundaries.
Warning: Because PWXPC can commit the change data to the targets between UOW boundaries, relational
integrity (RI) might be compromised. Do not use this connection attribute if you have targets in the CDC
session with RI constraints.
The maximum rows limit is cumulative across all sources in the CDC session. PWXPC issues a real-time flush
when the limit value is reached, regardless of the number of sources to which the changes were originally
made.
Use a maximum rows limit when extremely large UOWs in the change stream might cause locking issues on
the target database or resource issues on the node running the PowerCenter Integration Service. When you
specify a low maximum rows limit, the session consumes more system resources on the PowerCenter
Integration Service and target systems because PWXPC flushes data to the targets more frequently.
For example, a UOW contains 900 changes for one source followed by 100 changes for a second source and
then 500 changes for the first source. If you set the maximum rows value to 1000, PWXPC issues the commit
after reading 1,000 change records. In this example, the commit occurs after PWXPC processes the 100
changes for the second source.
Minimum Rows Per commit
For real-time or continuous extraction mode, minimum number of change records that PowerExchange reads
from the change stream before it passes a commit record to PWXPC. Until the minimum rows limit is met,
PowerExchange discards any commit records that it reads from the change stream and passes only change
records to PWXPC. After the minimum rows limit is met, PowerExchange passes the next commit record to
PWXPC and then resets the minimum rows counter.
Default is 0, which means that PowerExchange does not use minimum rows.

If you specify a minimum rows limit, PowerExchange changes the number of change records in a UOW to
match or exceed the limit. PWXPC does not commit change data to the targets when the minimum rows limit
occurs. PWXPC only commits change data to the targets based on the values of the
Maximum Rows Per commit, Real-Time Flush Latency in milli-seconds, and UOW Count attributes.
A minimum rows limit does not impact the relational integrity of the change data because PowerExchange
does not create new commits points in the change stream data. It merely skips some of the original commit
records in the change stream.
If your change data has many small UOWs, you can set the Minimum Rows Per commit attribute to create
larger UOWs of a more uniform size. Online transactions that run in transaction control systems such as CICS
and IMS often commit after making only a few changes, which results in many, small UOWs in the change
stream. PowerExchange and PWXPC process fewer, larger UOWs more efficiently than many small UOWs.
By using the minimum rows limit to increase the size of UOWs, you can improve CDC processing efficiency.
Real-Time Flush Latency in milli-seconds
For real-time or continuous extraction mode, number of milliseconds that must pass before PWXPC flushes
the data buffer to commit the change data to the targets. After the flush latency interval expires and PWXPC
reaches a UOW boundary, PWXPC issues a real-time flush to commit the change data and the restart tokens
to the targets and writes the PWXPC_10082 message in the session log. PWXPC resets the flush latency
interval when a real-time flush occurs because either the interval expires, or one of the UOW count or
maximum row limit is met.
Enter one of the following values for the flush latency interval:
¨ -1. Disables data flushes based on time.
¨ 0 to 2000. Interval set to 2000 milliseconds, or 2 seconds.
¨ 2000 to 86400.Interval set to the specified value.
Default is 0, which means that PWXPC uses 2,000 milliseconds.
If you set the flush latency interval value is 0 or higher, PWXPC flushes the change data for all complete
UOWs after the interval expires and the next UOW boundary occurs. The lower you set the flush latency
interval value, the faster you commit change data to the targets. Therefore, if you require the lowest possible
latency for the apply of changes to the targets, specify a low value for the flush latency interval.
When you specify low flush latency intervals, the CDC session might consume more system resources on the
PowerCenter Integration Service and target systems because PWXPC commits to the targets more
frequently. When you choose the flush latency interval value, you must balance performance and resource
consumption with latency requirements.
UOW Count
Number of complete UOWs that PWXPC reads from the change stream before flushing the change data to the
targets. As PWXPC reads change data from PowerExchange and provides that data to the appropriate source
in the CDC session, it counts the number of UOWs. After the UOW count value is reached, PWXPC issues a
real-time flush to commit the change data and the restart tokens to the targets, and writes the PWXPC_10081
message in the session log. PWXPC resets the UOW count when a real-time flush occurs because the UOW
count or maximum rows limit is met, or the flush latency interval expires.
Enter one of the following for the UOW count value:
¨ -1 or 0. PWXPC does not use the UOW Count attribute to control commit processing.
¨ 1 to 999999999. PWXPC flushes change data after reading the number of UOWs specified by
UOW Count attribute.
Default is 1.

The lower you set the value for the UOW Count attribute, the faster that PWXPC flushes change data to the
targets. To achieve the lowest possible latency for applying change data to targets, set the UOW Count
attribute to 1. However, the lowest possible latency for applying change data also results in the highest
possible resource consumption on the PowerCenter Integration Service and the target systems.
Commit processing for CDC sessions is not controlled by a single commitment control attribute. The
Maximum Rows Per commit, Real-Time Flush Latency in milli-seconds, and UOW Count values all result in a
real-time flush of change data, which causes the data and restart tokens to be committed to the targets. When you
choose values for the UOW Count, Real-Time Flush Latency in milli-seconds, and
Maximum Rows Per commit attributes, balance performance and resource consumption with latency
requirements.
Warning: You must ensure that the session properties Commit Type attribute specifies Source and that the
Commit at End of File attribute is disabled. By default, the Commit at End of File attribute is enabled, which
causes the PowerCenter Integration Service to write additional data to the targets after the CDC reader has
committed the restart tokens and shut down. As a result, when you restart the CDC session, duplicate data might
be written to the targets.
For more information, see “Commit Processing with PWXPC” on page 118.
Creating Restart Tokens for Extractions

Before you extract change data, you must establish an extraction start point. An optimal extraction start point
matches a time in the change stream that occurs after the target has been synchronized with the source but before
any new changes occur for the source. Usually, this point is the end of the change stream because changes to the
source are inhibited until the target is materialized and restart tokens are generated.
You can generate current restart tokens for the end of the change stream by using one the following methods:
¨ PWXPC restart token file. Generate current restart tokens for CDC sessions that use real-time or continuous
extraction mode by coding the CURRENT_RESTART option on the RESTART1 and RESTART2 special
override statements in the PWXPC restart token file. When the session executes, PWXPC requests that
PowerExchange provide restart tokens for the current end of the change stream, which PWXPC then uses as
the extraction start point.
¨ Database Row Test. Generate current restart tokens for sources by performing a database row test in
PowerExchange Navigator and coding a SELECT CURRENT_RESTART SQL statement.
¨ DTLUAPPL utility. Generate current restart tokens for sources by using the GENERATE RSTKKN option in
the DTLUAPPL utility.
If you use a PowerExchange utility or the PowerExchange Navigator to generate restart tokens, edit the restart
token file that PWXPC uses to specify the token values before you start the CDC session.
Displaying Restart Tokens

In the PowerExchange Navigator, you can perform a database row test on an extraction map to display the restart
token pair for each row of change data. The database row test output includes the following columns for the token
values:
¨ DTL__CAPXRESTART1 column for the sequence token
¨ DTL__CAPXRESTART2 column for the restart token
Creating Restart Tokens for Extractions 135

If you include the DTL__CAPXRESTART1 and DTL__CAPXRESTART2 columns in your PowerCenter source
definition, PowerExchange provides the restart tokens for each row when you extract change data in a CDC
session.
When a CDC session runs, PowerExchange and PWXPC display restart token values in the following messages:
¨ In the messages PWX-04565 and PWX-09959, the sequence token is in the Sequence field and restart token is
in the PowerExchange Logger field.
¨ In the messages PWXPC_12060 and PWXPC_12068, the sequence token is in the Restart Token 1 field and
the restart token is in the Restart Token 2 field.
¨ In the messages PWXPC_10081, PWXPC_10082, and PWXPC_12128, the sequence token is the first token
value and is followed by the restart token.
When you use the DTLUAPPL utility to generate restart tokens, use the PRINT statement to display the generated
values. In the PRINT output, DTLUAPPL displays the sequence token, without the usual trailing eight zeros, in the
Sequence field and displays the restart token in the Restart field.
Configuring the Restart Token File

When you configure the CDC session in PowerCenter, specify the name and location of the restart token file in the
following attributes of the source PWX CDC application connection:
¨ RestartToken File Folder. Specify the directory that contains the restart token file. If the folder does not exist
and the attribute contains the default value of $PMRootDir/Restart, PWXPC creates it. PWXPC does not create
any other restart token folder name.
¨ RestartToken File Name. Specify the unique name of the restart token file. If you do not specify a value in this
attribute, PWXPC uses the value of the Application Name, if available. Otherwise, PWXPC uses the name of
the workflow. Because this name must be unique, Informatica recommends that you always code a value for
the RestartToken File Name attribute.
When you run a CDC session, PWXPC verifies that the restart token file exists. If one does not exist, PWXPC
uses the name specified in the RestartToken File Name attribute to create an empty restart token file.
Restriction: The value of RestartToken File Name attribute in must be unique for every CDC session. Non-
unique file names can cause unpredictable results, such as change data loss and session failures.
To locate the restart token file name for a CDC session, check the following places:
¨ For existing CDC sessions, message PWXPC_12057 in the session log contains the restart token file folder
and the restart token file name.
¨ In Workflow Manager, the PWX CDC application connection associated with the source in the CDC session
contains the restart token file name and folder location. If the restart token file name is not specified in the
application connection, PWXPC uses the application name, if specified. Otherwise, PWXPC uses the workflow
name.
Before you run a CDC session for the first time, configure the restart token file to specify the point in the change
stream from which PowerExchange begins to extract change data. You can also configure the restart token file to
add new sources to a CDC session or to restart change data extraction from a specific point in the change stream.

Restart Token File Statements
You can use the following types of statements in a the restart token file:
¨ Comment
¨ Explicit override. Specify a restart token pair for a specific source. You must provide the PowerExchange
extraction map name.
¨ Special override. Specify a restart token pair for one or more sources. You can provide a specific restart token
pair or request that PowerExchange use the current restart point.
Restart Token File Statement Syntax

For the comment statements, use the following syntax:

restart1=000000AD775600000000000000AD77560000000000000000
Restart2=C1E4E2D34040000000AD5F2C00000000

d1dsn9.rrtb0001_RRTB_SRC_001=0000060D1DB2000000000000060D1DB20000000000000000
d1dsn9.rrtb0001_RRTB_SRC_001=C1E4E2D340400000013FF36200000000
d1dsn9.rrtb0002_RRTB_SRC_002=000000A3719500000000000000A371950000000000000000
d1dsn9.rrtb0002_RRTB_SRC_002=C1E4E2D34040000000968FC600000000
d1dsn9.rrtb0004_RRTB_SRC_004=000006D84E7800000000000006D84E780000000000000000
d1dsn9.rrtb0004_RRTB_SRC_004=C1E4E2D340400000060D1E6100000000
When you warm start the CDC session, PWXPC reads the restart token file to process any override statements for
restart tokens. In this case, the restart token file overrides all restart tokens for all sources in the CDC session.
After resolving the restart tokens for all sources, PWXPC writes message PWXPC_12060 to the session log with
the following information:
===============================
Session restart information:
===============================
Extraction Map Name Restart Token 1 Restart Token 2 Source
d1dsn9.rrtb0001_RRTB_SRC_001 0000060D1DB2000000000000060D1DB20000000000000000 C1E4E2D340400000013FF36200000000 Restart file
d1dsn9.rrtb0002_RRTB_SRC_002 000000A3719500000000000000A371950000000000000000 C1E4E2D34040000000968FC600000000 Restart file
d1dsn9.rrtb0003_RRTB_SRC_003 000000AD775600000000000000AD77560000000000000000 C1E4E2D34040000000AD5F2C00000000 Restart file (special
override)
d1dsn9.rrtb0004_RRTB_SRC_004 000006D84E7800000000000006D84E780000000000000000 C1E4E2D340400000060D1E6100000000 Restart file
override)
override)
override)
PWXPC indicates the source of the restart token values for each source. For the sources that had explicit override
statements in the restart token file, PWXPC writes “Restart file” in the Source column.
For the sources to which PWXPC assigns the special override restart tokens, PWXPC writes “Restart file (special
override)” in the Source column.
Configuring the Restart Token File 139

CHAPTER 9
Managing Change Data Extractions

¨ Starting PowerCenter CDC Sessions, 140

¨ Stopping PowerCenter CDC Sessions, 142
¨ Changing PowerCenter CDC Sessions, 144
¨ Recovering PowerCenter CDC Sessions, 146
Starting PowerCenter CDC Sessions

Use Workflow Manager, Workflow Monitor, or pmcmd to start a workflow or task for a CDC session. You can start
the entire workflow, part of a workflow, or a task in the workflow. You can do a cold start, warm start, or recovery
start. The method you use determines how PWXPC acquires the restart information.
Use one of the following methods to start a CDC session:
Cold start
To cold start a CDC session, use the Cold Start command in Workflow Manager or Workflow Monitor. You can
also use the pmcmd starttask or startworkflow commands with the norecovery option. A CDC session that
uses real-time or continuous extraction mode runs continuously until it is stopped or interrupted. A CDC
session that uses batch extraction mode runs until it reaches the end of log (EOL) or it is stopped or
interrupted.
When you cold start a CDC session, PWXPC uses the restart token file to acquire restart tokens for all
sources. PWXPC does not read the state tables or file or makes any attempt to recover the session.
Warm start
To warm start a CDC session, use the Start or Restart commands in Workflow Manager or Workflow Monitor.
You can also use the pmcmd starttask or startworkflow commands. A CDC session that uses real-time or
extraction mode runs continuously until it is stopped or interrupted. A CDC session that uses batch extraction
mode runs until it reaches EOL or it is stopped or interrupted.
When you warm start a CDC session, PWXPC reconciles any restart tokens provided in the restart token file
with any restart tokens that exist in the state tables or file. If necessary, PWXPC performs recovery
processing.
Recovery start
To start recovery for a CDC session, use the Recover command from Workflow Manager or Workflow Monitor.
You can also use the pmcmd recoverworkflow command or the starttask or startworkflow commands with the
recovery option. When recovery completes, the CDC session ends.
140
When you recover a CDC session, PWXPC reads the restart tokens from any applicable state tables or file. If
necessary, PWXPC performs recovery processing. PWXPC updates the restart token file with the restart
tokens for each source in the CDC session, and then the session ends. To begin extracting change data
again, either cold start or warm start the session.
Cold Start Processing

Cold start workflows and tasks by using the Cold Start command in Workflow Manager or Workflow Monitor. You
can also use the pmcmd starttask or startworkflow commands with the norecovery option.
After you request a cold start for a CDC session, the following processing occurs:
1. PWXPC writes the following message in the session log:

PWXPC_12091 [INFO] [CDCRestart] Cold start requested
2. PWXPC reads the restart tokens from only the restart token file and associates a restart token with each
source in the session.
3. PWXPC creates the initialization restart token file with the initial restart tokens.
4. PWXPC commits the restart tokens for each source to the appropriate state tables or file and then writes the
message PWXPC_12104 to the session log.
5. PWXPC passes the restart tokens to PowerExchange. PowerExchange begins extracting change data and
passing the data to PWXPC for processing.
6. PWXPC continues processing change data from PowerExchange and commits the data and restart tokens to
the targets. This processing continues until the session ends or is stopped.
Warm Start Processing

Warm start workflows and tasks by using the Start or Restart command in Workflow Manager or Workflow Monitor.
You can also use the pmcmd starttask or startworkflow commands.
When you warm start a workflow or task, PWXPC automatically performs recovery. You do not need to recover
failed workflows and tasks before you restart them.
After you request a warm start for a CDC session, the following processing occurs:

PWXPC_12092 [INFO] [CDCRestart] Warm start requested. Targets will be resynchronized automatically
if required
2. PWXPC queries the PowerCenter Integration Service about the commit levels of all targets. If all targets in the
session have the same commit level, PWXPC skips recovery processing.
3. PWXPC reconciles the restart tokens from the restart token file and from the state tables or file.
Restriction: If a CDC session requires recovery processing, PWXPC does not use the restart token file.
Consequently, you cannot override restart tokens for sources.
4. PWXPC creates the initialization restart token file with the reconciled restart tokens.
5. If recovery is required, PWXPC re-reads the change data for the last unit-of-work (UOW) that was committed
to the targets with the highest commit level and flushes the data to those targets with lower commit levels.
The PowerCenter Integration Service commits flushed change data and restart tokens to any relational
targets and updates any nonrelational files.
6. If recovery is not required and the reconciled restart tokens differ from those in the state tables or file,
PWXPC commits the reconciled restart tokens and then writes message PWXPC_12104 to the session log.
Starting PowerCenter CDC Sessions 141

7. PWXPC passes the restart tokens to PowerExchange. PowerExchange begins extracting change data and
passing the data to PWXPC for processing.
8. PWXPC continues processing change data from PowerExchange and commits the data and restart tokens to
the targets. This processing continues until the session ends or is stopped.
Recovery Processing
Recover workflows and tasks by selecting the Recover command in Workflow Manager or Workflow Monitor. You
can also use the pmcmd recoverworkflow command, or the starttask or startworkflow command with the recovery
option.
You can use recovery to populate the restart token file with the restart tokens for all sources in a CDC session so
that you can then cold start the CDC session or to ensure that the targets and restart tokens are in a consistent
state. However, you do not need to recover failed workflows and tasks before you restart them because PWXPC
automatically performs recovery processing when you warm start a workflow or task.
After you request recovery for a CDC session, the following processing occurs:

PWXPC_12093 [INFO] [CDCRestart] Recovery run requested. Targets will be resynchronized if required
and processing will terminate
2. PWXPC queries the PowerCenter Integration Service about the commit levels of all targets. If all targets in the
session have the same commit level, PWXPC skips recovery processing.
3. PWXPC reads the restart tokens from the recovery state tables or file.
Restriction: If a CDC session requires recovery processing, PWXPC does not use the restart token file.
Consequently, you cannot override restart tokens for sources.
4. PWXPC creates the initialization restart token file with the reconciled restart tokens.
5. If recovery is required, PWXPC re-reads the change data for the last UOW that was committed to the targets
with the highest commit level and flushes the data to those targets with lower commit levels. The
PowerCenter Integration Service commits any flushed change data and restart tokens to any relational
targets, and updates any nonrelational files.
6. PWXPC updates the restart token file with the final restart tokens, creates the termination restart token file,
and ends.
To process change data from the point of recovery, warm start or cold start the workflow or task.
Stopping PowerCenter CDC Sessions

You can stop CDC sessions from PowerCenter or PowerExchange. In PowerCenter, issue the Stop or Abort
command in Workflow Monitor. You can also use pmcmd stoptask, stopworkflow, aborttask, or abortworkflow
commands. In PowerExchange, issue the STOPTASK command or run the DTLUTSK utility.
Use one of the following methods to stop a running CDC session:
Stop
Use the Stop command in Workflow Monitor or the pmcmd stoptask or stopworkflow commands. After the
PWXPC CDC reader and PowerCenter Integration Service process all of the data in the pipeline and shut
down, the session ends.
142 Chapter 9: Managing Change Data Extractions

STOPTASK
Use the PowerExchange STOPTASK command. You can run the STOPTASK command on the source system
that is extracting the change data, from the PowerExchange Navigator, or by using pwxcmd or the DTLUTSK
utility. When you issue the STOPTASK command, PowerExchange stops the extraction task in the
PowerExchange Listener and passes an EOF to the PowerCenter Integration Service, which ends the session.
Abort
Use the Abort command in Workflow Monitor or the pmcmd aborttask or abortworkflow commands. When you
abort a CDC session, the PowerCenter Integration Service waits 60 seconds to allow the readers and the
writers time to process all of the data in the pipeline and shut down. If the PowerCenter Integration Service
cannot finish processing and committing data within this timeout period, it kills the DTM process and ends the
session.
Stop Command Processing

Stop CDC sessions and workflows by using the Stop command in Workflow Monitor or the pmcmd stopttask or
stopworkflow command. You can also use the PowerExchange STOPTASK command.
After you issue a stop command in PowerCenter or PowerExchange, the following processing occurs:
1. If you use a PowerCenter stop command, the PowerCenter Integration Service requests PWXPC to stop.
If you use a PowerExchange stop command, PowerExchange sends an EOF to PWXPC.
2. When PWXPC receives an EOF, it flushes any complete and uncommitted UOWs with the associated restart
tokens to the targets. PWXPC then writes the messages PWXPC_12101 and PWXPC_12068 to the session
log.
3. The PowerCenter Integration Service processes all of data in the pipeline and writes it to the targets.
4. The PowerCenter Integration Service sends an acknowledgment to PWXPC indicating that the targets have
been updated.
5. PWXPC writes the termination restart token file, and then writes the message PWXPC_12075 to the session
log.
6. The PWXPC CDC reader shuts down.
7. The PowerCenter Integration Service performs any post-session tasks and ends the session.
Terminating Conditions
To stop a CDC session based on a user-defined event or at EOL, configure a termination condition in the session.
A terminating condition determines when the PWXPC stops reading change data from the sources and ends the
CDC session. After PWXPC reaches a terminating condition, it flushes the change data to the targets and passes
an EOF to the PowerCenter Integration Service. The PowerCenter Integration Service commits the data to the
targets and ends the session.
You can configure the following termination conditions for CDC sessions:
¨ Event table processing. If you specify an extraction map table in the Event Table attribute of the PWX CDC
Real Time application connection, PowerExchange, after it reads a change record for the event table, passes
EOF to PWXPC to end the CDC session.
¨ Idle Time. If you specify 0 for the Idle Time attribute on a PWX CDC Real Time application connection,
PowerExchange, after it reaches EOL, passes EOF to PWXPC to end the CDC session.
¨ Batch extraction mode. If you use batch extraction mode by configuring a PWX CDC Change application
connection, PowerExchange, after it reads all closed PowerExchange Condense condense files or
PowerExchange Logger for Linux, UNIX, and Windows log files, passes PWXPC EOF to end the CDC session.
Stopping PowerCenter CDC Sessions 143

Changing PowerCenter CDC Sessions
You can add new sources and targets to an existing CDC sessions. Afterward, you must cold start the session.
Because a cold start is required, you must also get the latest restart tokens for the original sources prior to
restarting the session. To do so, you can perform a recovery.
To change a PowerCenter CDC session:
1. Stop the workflow.

2. After the workflow ends, recover the CDC session.
When you recover tasks, PWXPC writes the ending restart tokens for all sources in a CDC session to the
restart token file that you specified on the PWX CDC application connection.
3. Make changes to the session or workflow, if necessary.
4. Verify that the restart token file in the source CDC connection points to the same restart token file updated in
the recovery.
5. If you add sources to the CDC session, add statements to the restart token file that provide restart tokens for
the new sources.
6. If you remove sources from the CDC session, update the restart token file to remove their restart tokens.
7. Cold start the CDC session.
Examples of Creating a Restart Point

The following examples show different methods of creating a restart point for a source table that is added to an
existing CDC session. The first example uses the CURRENT_RESTART option of the special override statement
in the restart token file to generate current restart tokens. The second example uses DTLUAPPL to generate
current restart tokens.
Adding a New Source and Use CURRENT_RESTART to Create Restart Tokens

- Example
In this example, a new source table, RRTB_SRC_004, is added to an existing CDC session that contains three
sources. The restart points for the existing sources are maintained. For the new source, the example uses the
CURRENT_RESTART option in the restart token file to generate a restart token that represents the current end of
the change stream.
To add a new source and use CURRENT_RESTART to create restart tokens:
1. To stop the workflow, select the Stop command in Workflow Monitor.

2. After the workflow stops, select the Recover Task command in Workflow Monitor to run a recovery session.
PWXPC writes the following messages in the session log:
PWXPC_12060 [INFO] [CDCRestart]
===============================
===============================
d1dsn9.rrtb0002_RRTB_SRC_002 000000AD220F00000000000000AD220F0000000000000000 C1E4E2D34040000000AD0D9C00000000 GMD storage
PWXPC also writes the restart tokens in the restart token file specified in the CDC application connection.
3. Edit the mapping, session, and workflow to add the new source, RRTB_SRC_004.
4. Edit the restart token file to specify the CURRENT_RESTART option for the new source.

The updated file appears as follows:
<!-- existing sources
d1dsn9.rrtb0001_RRTB_SRC_001=000000AD220F00000000000000AD220F0000000000000000
d1dsn9.rrtb0001_RRTB_SRC_001=C1E4E2D34040000000AD0D9C00000000
<!-- new source
RESTART1=CURRENT_RESTART
RESTART2=CURRENT_RESTART
5. Cold start the session.
PWXPC connects to PowerExchange and generates restart tokens that match the current end of the change
stream for the new source, RRTB_SRC_004. PWXPC then passes the restart tokens to PowerExchange to
begin change data extraction. Because the restart points for the other sources are earlier than the one just
generated for RRTB_SRC_004, PWXPC does not pass any change data to this new source until the first
change following its generated restart point is read.
Adding a New Source and Use DTLUAPPL to Create Restart Tokens - Example
In this example, a new source table, RRTB_SRC_004, is added to an existing CDC session containing three
sources. The restart points for the existing sources are maintained. The DTLUAPPL utility is used to generate a
restart token that represent the current end of the change stream.
1. To stop the workflow, select the Stop command in Workflow Monitor.

2. After the workflow stops, select the Recover Task command from Workflow Monitor to run a recovery session.
PWXPC writes the following messages in the session log:
===============================
===============================
PWXPC also writes the restart tokens in the restart token file specified in the CDC application connection.
3. Edit the mapping, session, and workflow to add the new source, RRTB_SRC_004.
4. Run DTLUAPPL with RSTTKN GENERATE to generate restart tokens for the current end of the change
stream. Use the following DTLUAPPL control cards:
mod APPL dummy DSN7 rsttkn generate
mod rsttkn rrtb004
end appl dummy
print appl dummy
The PRINT command produces the following output:
Registration name=<rrtb004.1> tag=<DB2DSN7rrtb0041>
Sequence=<00000DBF240A0000000000000DBF240A00000000>
Restart =<C1E4E2D3404000000DBF238200000000>
Add eight zeros to the end of the Sequence value to create the sequence value for the restart token file.
5. Edit the restart token file to add the new source and its tokens.
The updated file contains the following lines:
<!-- existing sources
<!-- new source
Changing PowerCenter CDC Sessions 145

d1dsn9.rrtb0004_RRTB_SRC_004=00000DBF240A0000000000000DBF240A0000000000000000
d1dsn9.rrtb0004_RRTB_SRC_004=C1E4E2D3404000000DBF238200000000
6. Cold start the session.
PWXPC passes these restart tokens to PowerExchange to begin change data extraction. Because the restart
points for the other sources are earlier than the one just generated for RRTB_SRC_004, PWXPC does not
pass any change data to this new source until the first change following the generated restart point is read.
Recovering PowerCenter CDC Sessions

Use Workflow Manager, Workflow Monitor, or pmcmd to recover a workflow or task for a CDC session that fails.
You can recover the entire workflow or a task in the workflow.
A CDC session can fail for the following reasons:
¨ Permanent errors, such as source or target data errors
¨ Transitory or environmental errors, such as infrastructure problems, server failures, and network availability
issues
If you run a session with a resume recovery strategy and the session fails, do not edit the state information or the
mapping for the session before you restart the session.
If a session fails because of transitory or environmental errors, restart the session after you have corrected the
errors. When you warm start a CDC session, PWXPC automatically performs recovery, if required. Alternatively,
you can recover a CDC session, and then restart the session.
If a CDC session fails because of permanent errors, such as SQL or other database errors, you must correct the
errors before restarting the CDC session. With some failures, you can correct the error and then restart the CDC
session. In other cases, you might need to rematerialize the target table from the source table before you start
extracting and applying change data again. If you rematerialize the target table, you should provide restart tokens
that match the materialization point in the change stream, and then cold start the CDC session.
Restriction: If a CDC session requires recovery processing, you cannot override the restart tokens because
PWXPC does not read the restart token file.
Example of Session Recovery

In this example, a CDC session with relational targets is aborted in the Workflow Monitor. Then, the Restart Task
command is issued from the Workflow Monitor to restart the CDC session.
When you warm start the session, PWXPC automatically performs a recovery, and writes the following message in
the session log:
PWXPC_12092 [INFO] [CDCRestart] Warm start requested. Targets will be resynchronized automatically if
required
PWXPC then reads the restart tokens from the state tables or file and writes the message PWXPC_12060 in the
session log. The PWXPC_12060 message records the restart tokens for the session and its sources, as shown in
the following example:
===============================
===============================
d1dsn8.rrtb0004_RRTB_SRC_004 00000FCA65840000000000000D2E004A00000000FFFFFFFF C1E4E2D3404000000D21B1A500000000 GMD storage

If PWXPC detects that recovery is required, PWXPC writes the message PWXPC_12069 in the session log. This
message usually includes the restart tokens for both the begin-UOW and the end-UOW for the oldest uncommitted
UOW that PWXPC re-reads during recovery. PWXPC usually stores end-UOW restart tokens in the state table or
file. However, if you specify a maximum rows threshold, PWXPC can commit change data and restart tokens
between UOW boundaries. As a result, the restart tokens might not represent an end-UOW.
The following example PWXPC_12069 message include “from” restart tokens that are the same as those
displayed in the example PWXPC_12060 message:
PWXPC_12069 [INFO] [CDCRestart] Running in recovery mode. Reader will resend the the oldest uncommitted UOW to resync targets:
from: Restart 1 [00000FCA65840000000000000D2E004A00000000FFFFFFFF] : Restart 2 [C1E4E2D3404000000D21B1A500000000]
to: Restart 1 [00000FCA65840000000000000D300D8000000000FFFFFFFF] : Restart 2 [C1E4E2D3404000000D21B1A500000000].
Because this session specifies a maximum rows threshold, the restart token values in the Restart 2 fields in both
the “from” and “to” restart tokens is the begin-UOW value. The sequence token values in the Restart 1 fields
represent the start and end change records in the UOW that is displayed in the Restart 2 field.
During recovery processing, PWXPC reads the change data records between the points defined by the two restart
token values in the PWXPC_12069 message and then issues a commit for the data and the restart tokens. The
PowerCenter Integration Service writes the flushed change data to the target tables and writes the restart tokens
to the state table. Then the session ends.
Recovering PowerCenter CDC Sessions 147

CHAPTER 10
Monitoring and Tuning Options

¨ Monitoring Change Data Extractions, 148

¨ Tuning Change Data Extractions, 154
¨ CDC Offload and Multithreaded Processing, 159
Monitoring Change Data Extractions

PowerExchange, PWXPC, and PowerCenter issue messages that you can use to monitor the progress of CDC
sessions. PWXPC can also display progress and statistical information about CDC sessions in the PowerCenter
Workflow Monitor.
Monitoring CDC Sessions in PowerExchange

In PowerExchange, you can use the following information to monitor the extraction of change data by CDC
sessions:
¨ Read progress messages. You can request that PowerExchange write messages that indicate the number of
change records read by a CDC session.
¨ Extraction statistics messages. When extraction sessions end, PowerExchange writes messages that
include statistical information about the change records processed.
¨ Multithreaded processing statistics messages. You can request that PowerExchange write statistical
information about CDC sessions that use multithreaded processing.
¨ LISTTASK command output. You can use the LISTTASK command to display active CDC sessions.
Read Progress Messages

You can request that PowerExchange write messages that indicate read progress to the PowerExchange log file.
If you select the Retrieve PWX log entries option on a PWX CDC application connection, PWXPC writes the
progress messages in the session log.
To direct PowerExchange to write read progress messages, include the following parameters in the DBMOVER
configuration file:
¨ PRGIND. Specify Y to have PowerExchange write PWX-04587 messages that indicate the number of records
read for a CDC session. Default is N.
¨ PRGINT. Specify the number of records that PowerExchange reads before writing the PWX-04587 messages
to the PowerExchange log file. Default is 250 records.
148
The PWX-04587 messages have the following format:
PWX-04587 int_server/workflow_name/session_name: Records read=num_records
Where:
¨ int_server is the name of the PowerCenter Integration Service.
¨ workflow_name is the name of the workflow that contains the CDC session.
¨ session_name is the name of the CDC session.
¨ num_records is the cumulative number of records read since the CDC session started.
For example, to direct PowerExchange to write read progress messages after 100 records, the DBMOVER
configuration file contains the following parameters:
PRGIND=Y
PRGINT=100
When a CDC session that has a session name of s_cdc_DB2_SQL_stats runs, PowerExchange writes the
following messages to the PowerExchange log file:
PWX-04587 intserv/wf_cdc_mon_stats/s_cdc_DB2_SQL_stats: Records read=100
PowerExchange continues to write PWX-04587 messages for this CDC session until the session ends. In the
PowerExchange log file, each of these messages has a date and timestamp. You can use this information to
determine the speed with which PowerExchange processes change data from the change stream.
Extraction Statistics Messages

When a CDC session ends, PowerExchange writes the following messages that contain statistical information
about the session:
¨ PWX-04578. PowerExchange writes this message for each source in the CDC session. This message includes
the number of insert, update, delete, commit, and total records read for the source.
¨ PWX-04588. PowerExchange writes this message for the entire CDC session. This message includes the total
number of records read for that CDC session.
Important: The statistical information in the PowerExchange messages represents the change data that
PowerExchange read for a CDC session. This information might not reflect the data that was applied to the
targets. For statistical information about the change data applied to the target, review the session log.
Multithreaded Processing Statistics

If you use CDC offload processing, you can also use multithreaded processing to attempt to increase throughput
on the PowerCenter Integration Service machine where the offloaded processing runs.
To monitor the effectiveness of multithreaded processing, specify the following parameter in the DBMOVER
configuration file on the PowerCenter Integration Service machine:
SHOW_THREAD_PERF=number_records
Number of change records that PowerExchange reads during a statistics reporting interval before writing the
statistics messages PWX-31524 through PWX-31259 to the PowerExchange log file. If you select the
Retrieve PWX log entries option on the connection in the CDC session, PWXPC writes these messages in
the session log.
You can use the information in the messages to tune multithreaded processing. For PowerExchange to write
statistics messages for threads, you must specify 1 or above for Worker Threads on the connection.
Otherwise, PowerExchange does not use multithreaded processing or produce statistics messages.
Monitoring Change Data Extractions 149

The messages that PowerExchange writes during each statistics interval contain the following information:
¨ PWX-31255. Cycle time, which is the total time that PowerExchange on the PowerCenter Integration Service
machine spent processing the change data before passing it to PWXPC. This message includes the total
percentage of time and average, minimum, and maximum times in microseconds.
¨ PWX-31256. I/O time, which is the time that PowerExchange on the PowerCenter Integration Service machine
spent reading change data from the PowerExchange Listener on the source system. This message includes the
I/O percentage of the total time and average, minimum, and maximum times in microseconds.
¨ PWX-31257. Parsing time, which is the time that PowerExchange on the PowerCenter Integration Service
machine spent in column-level processing for the change records on all threads. This message includes the
parsing percentage of the total time and average, minimum, and maximum times in microseconds.
¨ PWX-31258. External time, which is the time that PowerExchange on the PowerCenter Integration Service
machine spent combining the change records from all threads back into a single UOW to pass to PWXPC and
for PWXPC to flush the data to PowerCenter. This message includes the external percentage of the total time
and average, minimum, and maximum times in microseconds.
¨ PWX-31259. Delay time, which is the time that the PowerExchange on the PowerCenter Integration Service
machine waited to receive new change records to process from the PowerExchange Listener on the source
system. This message includes the delay percentage of the total time and average, minimum, and maximum
times in microseconds.
If the parsing and external processing times are higher than the I/O time, you might improve throughput by
increasing the number of threads for the CDC session.
For the following example, SHOW_THREAD_PERF=10000 is specified in the DBMOVER configuration file.
PowerExchange writes the following sample messages after 10,000 change records have been read and the next
UOW boundary is reached:
PWX-31254 PowerExchange threading stats for last 10000 rows. Cycle (array) size is 25 rows. 0 out of
array occured.
PWX-31255 Cycle time: 100% (avg: 5709 min: 4741 max: 7996 usecs)
PWX-31256 IO time: 4% (avg: 235 min: 51 max: 1021 usecs)
PWX-31257 Parse time: 79% (avg: 4551 min: 4102 max: 5495 usecs)
PWX-31258 Extern time: 20% (avg: 1145 min: 618 max: 3287 usecs)
PWX-31259 Delay time: 0% (avg: 7 min: 4 max: 165 usecs)
PWX-31254 PowerExchange threading stats for last 100000 rows. Cycle (array) size is 25 rows. 0 out of
array occured.
PWX-31255 Cycle time: 99% (avg: 5706 min: 4735 max: 7790 usecs)
PWX-31256 IO time: 4% (avg: 234 min: 51 max: 950 usecs)
PWX-31257 Parse time: 79% (avg: 4549 min: 4108 max: 5425 usecs)
PWX-31258 Extern time: 20% (avg: 1144 min: 616 max: 3242 usecs)
PWX-31259 Delay time: 0% (avg: 7 min: 4 max: 115 usecs)
DISPLAY ACTIVE or LISTTASK Command Output

Issue the PowerExchange Listener DISPLAY ACTIVE command to display CDC sessions that are active in the
PowerExchange Listener.
You can issue the command from the command line. On Windows, if you want to issue the command from the
PowerExchange Navigator, enter the equivalent LISTTASK command in the Database Row Test dialog box.
Alternatively, issue the pwxcmd listtask command from a Linux, UNIX, or Windows system to a PowerExchange
Listener running on the local system or a remote system.
The command output includes the PwrCntrSess field. This field provides the PowerCenter session name in the
following format:
integration_server_name/workflow_name/session_name
For example, if two active CDC sessions are active, the command produces the following output:
PWX-00711 Active tasks:
PWX-00712 TaskId=1, Partner=10.10.10.01, Port=2480, PwrCntrSess=intserv1/workflow1/cdc_sess1,
Application=appl_name1, Status=Active, AM=CAPXRT, Mode=Read, Process=, SessId=
150 Chapter 10: Monitoring and Tuning Options

PWX-00712 TaskId=2, Partner=10.10.10.02, Port=2480, PwrCntrSess=intserv2/workflow2/cdc_sess2,
Application=appl_name2, Status=Active, AM=CAPXRT, Mode=Read, Process=, SessId=
PWX-00713 2 active tasks
PWX-00709 0 Dormant TCBs
Monitoring CDC Sessions in PowerCenter

In PowerCenter, you can use the following information to monitor the progress of CDC sessions:
¨ Session log messages. PWXPC and PowerCenter write messages to the session log. You can use these
messages to monitor the progress of a CDC session.
¨ Performance details in Workflow Monitor. If you configure a CDC session to report performance details, you
can monitor the progress of the session in the Workflow Monitor.
Session Log Messages

You can use messages that PWXPC and PowerCenter write to the session log to monitor the progress of CDC
sessions.
When PWXPC flushes change data to commit the data to the targets, it writes one of the following messages to
the session log, displaying the reason for the flush:
PWXPC_10081 [INFO] [CDCDispatcher] raising real-time flush with restart tokens [restart1], [restart2]
because the UOW Count [count] is reached
because Real-time Flush Latency [latency] is reached
because the Maximum Rows Per commit [count] is reached
You can use the restart tokens in the PWXPC flush messages to monitor the processing of the change data. For
each PWXPC flush message, PowerCenter writes a WRT_8160 message after committing change data to the
targets. This messages displays the source-based commit statistics.
RELATED TOPICS:
¨ “Using Connection Options to Tune CDC Sessions ” on page 157
¨ “Tuning Commit Processing ” on page 159
¨ “Viewing Performance Details in the Workflow Monitor” on page 151
Viewing Performance Details in the Workflow Monitor

Performance details include counters that you can use to assess the efficiency of a CDC session and change data
extraction processing. The details include a single source qualifier that reflects group source processing for the
change data.
From Workflow Monitor, you can view the details for the current CDC session while it is executing. If you notice
degradation of CDC session performance, you can use the performance details to determine the bottleneck.
PWXPC does not store performance details in the repository so you cannot view previous performance details for
CDC sessions.
Note: To view performance details for a CDC session that has ended, you must select performance details while
the session is running. Otherwise, PWXPC does not display performance details.
To enable the collection of performance details, select Collect performance data on the Properties tab of the
CDC session. During the execution of the CDC session, PWXPC refreshes the statistical information every 10
seconds. If you have selected a resume recovery strategy in the CDC session, PWXPC displays data for all
performance counter fields.

To view performance details in the Workflow Monitor:
1. In Workflow Monitor, right-click a session and select Get Run Properties.

2. In the Properties window, click the Performance area.
The Performance Counter column displays a data source qualifier from the CDC session. The
Counter Value column displays the PowerCenter node name.
3. To view performance details, select the data source qualifier. The following table describes the fields that
PowerCenter displays in the Performance Counter column in the Performance area:
Performance Counter Field Description
1 PowerExchange CDC Reader Status: Current status of the PWXPC reader, as indicated by one of the
following values:
- No Data To Process. In the last read, PowerExchange did not
pass data to PWXPC.
- Restart Advance. PowerExchange passed restart tokens to
PWXPC but did not pass change data.
- Processing Data. PowerExchange passed change data and
restart tokens to PWXPC for processing.
1.1 Time Last Data Row Read Time, in milliseconds, when PWXPC last received data from
PowerExchange.
1.2 Data Rows In Current Interval Number of change records received from PowerExchange during
the current statistics interval.
1.3 End Packets In Current Interval Number of UOWs received from PowerExchange during the current
statistics interval.
1.4 Data Read Rate In Current Interval (rows/sec) Number of change records read per second by PowerExchange
during the current statistics interval.
The value varies, depending on the quantity of change data being
processed:
- If PowerExchange is reading large amounts of change data from
the change stream, this value is usually large and reflects the
maximum PowerExchange throughput.
- If PowerExchange is waiting for change data at the end of the
change stream, this value is small.
The following factors can increase this value:
- Large network bandwidth
- CDC offload processing
- Multithreaded processing
1.5 Mean Data Read Rate (rows/sec) Mean number of change records that PowerExchange read per
second, from the start of the CDC session.
1.6 Max Data Read Rate (rows/sec) Maximum number of change records that PowerExchange read per
second during a statistics interval, from the start of the CDC
session.

2 PowerCenter Processing Status: Overall status of the CDC session, as indicated by one of the
following values:
- Idle. Waiting for change data.
- Processing Data. Data is being processed.
- Recovery Disabled. If a resume recovery strategy is not
selected, the PWXPC CDC reader cannot obtain PowerCenter
status information.
2.1 Time Of Last Commit Timestamp of the last commit to a target.
2.2 Rows Processed To Commit In Current Interval Number of change records flushed by the PWXPC reader during
the current statistics interval. This count includes the change
records in all committed UOWs. Some of these UOWs might have
started before the current statistics interval began.
2.3 Commit Rate In Current Interval (rows/sec) Processing rate, in number of change records per second, for the
change records for the UOW that was last committed during the
current statistics interval. This rate includes reading the UOW from
PowerExchange and committing the change data to the targets.
The following factors can influence this rate:
- Number of available DTM buffers
- Responsiveness of the target
- Number of transformations in the pipeline
2.4 Mean Commit Rate (rows/sec) Mean number of change records per second for the rate displayed
in 2.3 Commit Rate In The Current Interval.
This value differs from the 2.6 Mean Throughput Rate in that it
takes into account only the time when the session is actively
processing data and does not reflect processing overlap in
PowerCenter.
2.5 Max Commit Rate (rows/sec) Maximum number of change records per second for the commit
rate displayed in 2.3 Commit Rate In The Current Interval,
recorded from the start of the CDC session.
2.6 Mean Throughput (rows/sec) Mean rate of processing for the CDC session.
2.7 Max Throughput (rows/sec) Maximum throughput for the CDC session.
2.8 Commits In Current Interval Number of commits processed to completion by the target during
the current statistics interval.
2.9 Commits Pending Number of commits that were issued by the PWXPC reader but that
have not yet reached the targets. A large value might indicate
problems with target responsiveness.
3 Capture Timestamps
3.1 Timestamp On Last End Packet Read The capture timestamp, DTL__CAPXTIMESTAMP, from the last
UOW read for a source in the CDC session.
3.2 Timestamp On Last Target Commit The capture timestamp, DTL__CAPXTIMESTAMP, from the last
UOW committed to the target.

4 Totals
4.1 Elapsed Time Total elapsed time for the CDC session.
4.2 Rows Read Total number of change records read from PowerExchange.
4.3 End Packets Read Total number of UOWs read.
4.4 Time in PowerExchange Processing Total time of PowerExchange processing for the CDC session.
4.5 Rows Processed Total number of change records processed through PowerCenter
and committed to the targets.
4.6 Commits to Target Total number of flushes that the PWXPC reader issued and that
were committed to the targets.
4.7 TS on Last Commit minus TS at Commit (2.1-3. Value that results from subtracting 3.2 Timestamp On Last Target
2) Commit value from the 2.1 Time Of Last Commit value. If this
result is negative, the value is enclosed in parentheses.
Tuning Change Data Extractions

You can use PowerExchange configuration parameters and connection options in PowerCenter to tune CDC
sessions. In addition, you can use CDC offload and multithreaded processing to improve throughput by moving
processing for change data to a different machine.
Use the following methods to tune CDC sessions:
¨ Parameters and options. To tune sessions, you can use specify parameters and options in the DBMOVER
configuration file and on PWX CDC connections.
¨ CDC offload processing. You can use CDC offload processing to distribute PowerExchange column-level
processing for change data to the PowerCenter Integration Service machine that runs the CDC session. By
distributing processing, you can reduce PowerExchange processing overhead on the system on which the
change data resides. You can also use CDC offload processing with the PowerExchange Logger for Linux,
UNIX, and Windows to capture change data on a different machine. CDC sessions can then extract change
data from the PowerExchange Logger log files on that machine, rather than from the change stream on the
original source machine.
¨ Multithreaded processing. If you use CDC offload processing, you can optionally use multithreaded
processing to attempt to increase throughput. Multithreaded processing uses multiple threads on the
PowerCenter Integration Service machine to perform the offloaded PowerExchange processing.
¨ Asynchronous network communication. PowerExchange uses asynchronous communication for most send
and receive operations, overlapping network processing with data processing. This feature is enabled
automatically and usually requires no tuning, but you can tune the feature if you need to.

Using PowerExchange Parameters to Tune CDC Sessions
To tune your PowerExchange installation, you can customize the following parameters in the DBMOVER
configuration file:
APPBUFSIZE=size
Defines the maximum size, in bytes, of the buffer that PowerExchange uses to read or write data. This data
buffer can exist on a source or target system.
If you are applying change data from the change stream on the source system to a remote target system,
PowerExchange usually writes change data to its application data buffer on the source system until the buffer
is full. PowerExchange then sends the data to a sending TCP/IP buffer on the source system. TCP/IP
transports the change data to a receiving TCP/IP buffer on the target system. PowerExchange on the target
system reads the change data from the TCP/IP buffer into its application data buffer. PWXPC then reads the
change data and passes it to PowerCenter. PowerCenter processes the data and applies it to the targets.
Enter an APPBUFSIZE value that is greater than the maximum size of any single data row to be sent.
Valid values are from 34816 through 1048576. Default is 128000.
If the target system is remote, enter the same APPBUFSIZE value in the DBMOVER configuration files on the
source and target systems. Also, verify that the APPBUFSIZE value matches the TCPIPBUFSIZE value in the
same DBMOVER configuration file. The TCPIPBUFSIZE parameter specifies the maximum size of the TCP/IP
buffer.
If the APPBUFSIZE value is not optimal, PowerExchange writes the PWX-01295 message in the
PowerExchange log file on the source system. This message includes a recommended minimum value.
COMPRESS={Y|N}
Defines whether PowerExchange uses its proprietary compression algorithm to compress data before it is
sent to TCP/IP for transmission to the remote platform.
Default is Y.
PowerExchange uses the COMPRESS setting in the DBMOVER configuration file on the remote system that
contacts the PowerExchange Listener. On the PWX CDC application connection, you can override the
compression setting in the DBMOVER configuration file. If you enable compression, the CPU consumption of
the PowerExchange Listener on the source system might increase.
To avoid unnecessary CPU consumption, set COMPRESS to N in the PowerExchange DBMOVER

configuration file on the PowerCenter Integration Service machine.
CAPI_CONNECTION=( ...,MEMCACHE=cache_value, ...))
Amount of memory cache, in kilobytes, that is allocated to reconstruct complete UOWs. You can specify the
MEMCACHE parameter on the following CAPI_CONNECTION statement types:
¨ MSQL
¨ UDB
¨ UOWC
PowerExchange keeps all changes in each UOW in cache until it processes the end-UOW record, which is
the commit record. If the MEMCACHE value is too small to hold all of the changes in a UOW in cache, the
changes spill to a disk file.
Valid values are from 1 through 519720. Default is 1024.
You might need to increase this value if you have large UOWs. PowerExchange processes a UOW more
efficiently if all of the changes are cached in memory. If a UOW might be larger than 1024 KB in size,
increase this parameter. For most environments, a value of 10240 (10 MBs) is a good starting value.
Tuning Change Data Extractions 155

Tip: PowerExchange uses the MEMCACHE value to allocate cache memory to each connection for change
data extractions. To prevent excessive memory use by a PowerExchange Listener, use a reasonable value
for MEMCACHE based on your extraction processing needs and the number of CDC sessions that run
concurrently.
CAPI_CONNECTION=( ...,RSTRADV=rstr_secs, ...))
Number of seconds that PowerExchange waits before advancing the restart tokens for a data source by
returning an empty unit of work (UOW). You can specify the RSTRADV parameter on the following
CAPI_CONNECTION statement types:
¨ MSQL
¨ UDB
¨ UOWC
Empty UOWs contain restart tokens only, without any data. PowerExchange uses the restart tokens to
determine the start point in the change stream for change data extractions. The wait period for the RSTRADV
value starts after a UOW for a data source is processed. PowerExchange resets the wait period after it reads
the next UOW for that source or when it returns an empty UOW because the wait period expires.
For sources with low change activity, you can use the RSTSADV parameter to periodically advance to the
restart tokens for those sources. Advancing the restart tokens speeds up restart processing for CDC sessions
by minimizing the amount of change data that must be reprocessed.
For example, if you specify RSTRADV=5 and changes are not made to the data source for five seconds,
PowerExchange returns an empty UOW to advance the restart point for the data source.
Valid values are from 0 through 86400. If you do not specify RSTRADV, PowerExchange does not return
empty UOWs to advance the restart point.
Consider the following issues when you set RSTRADV on CAPI_CONNECTION statements in the
PowerExchange DBMOVER configuration file:
¨ A value of 0 adversely affects performance. PowerExchange returns an empty UOW with restart tokens to
PWXPC after each UOW is processed.
¨ A low value can cause the UOW Count option on the PWX CDC connection to match more quickly than
expected. When the UOW counter matches, PWXPC flushes its data buffer and commits restart tokens to
the targets. Excessive flush activity can adversely affect performance on the PowerCenter Integration
Service machine and target databases.
LISTENER=(node_name,TCPIP,port,send_bufsize,receive_bufsize,send_msgsize,receive_msgsize, ...)
Defines a port on which a PowerExchange Listener listens for local or remote connections. The positional
parameters the send_bufsize, receive_bufsize, send_msgsize, and receive_msgsize define the send and
receive buffer and message sizes. If you do not specify values for these parameters, PowerExchange uses
the operating system defaults, which vary based on operating system.
To maximize throughput, consider increasing the send and receive buffer and message sizes on the
LISTENER statement on the source system. Contact your network administration to determine the best values
to use on your system.
Note: Do not specify values for the send and receive buffer and message sizes that exceed the TCP
maximum receive buffer size.
NODE=(node_name,TCPIP,hostname,port,send_bufsize,receive_bufsize,send_msgsize,receive_msgsize, ...)
Defines a port the IP information that PowerExchange uses to communicate with a remote PowerExchange
Listener. The positional parameters the send_bufsize, receive_bufsize, send_msgsize, and receive_msgsize
define the send and receive buffer and message sizes. If you do not specify values for these parameters,
PowerExchange uses the operating system defaults, which vary based on operating system.

To maximize throughput, consider increasing the send and receive buffer and message sizes on the NODE
statement on the target system. Contact your network administration to determine the best values to use on
your system.
Note: Do not specify values for the send and receive buffer and message sizes that exceed the TCP
maximum receive buffer size.
TRACE=(trace_id,trace_level,99)
Defines PowerExchange diagnostic traces that Informatica Global Customer Support uses to solve problems
with PowerExchange code.
TRACE statements can severely impact PowerExchange performance. You should use them only at the
direction of Informatica Global Customer Support. To enhance performance, remove or comment out all
TRACE statements in the DBMOVER configuration files on all systems.
RELATED TOPICS:
¨ “Using Connection Options to Tune CDC Sessions ” on page 157
Using Connection Options to Tune CDC Sessions

In PowerCenter, you can customize options on the PWX CDC connections to tune CDC sessions. The following
table describes the connection options that you can use to tune CDC sessions:
Connection Option Description Tuning Suggestion
Compression Select this option to compress source data Do not use compression.
during the PowerCenter session.
Default is disabled.
Encryption Type The type of data encryption that Do not use encryption.
PowerExchange uses.
Default is None.
Image Type Indicates whether PWXPC extracts after Set to AI.

images (AI) only or both before and after
images (BA) for change data.
If you use the PowerExchange Logger for
Linux, UNIX, and Windows and specified
CAPT_IMAGE=BA in the pwxccl.cfg
configuration file, you can set this option to AI
or BA. If you specify AI, before images of the
data can still be embedded in Update rows if
you add DTL_BI columns to the extraction
map. With DTL_BI columns, you can
manipulate before-image data in the mappings.
Default is BA.
UOW Count The number of UOWs that PWXPC reads from To improve efficiency on the PowerCenter
the source before it flushes the data buffer to Integration Service machine and the target
commit the change data to the targets. databases, reduce commit processing.
Default is 1.
Real-time Flush The frequency, in milliseconds, with which To improve efficiency on the PowerCenter
Latency in mill-seconds PWXPC flushes the data buffer to commit the Integration Service machine and the target
change data to the targets. databases, reduce commit processing.
Default is 0, which is equivalent to two seconds.
Tuning Change Data Extractions 157

Connection Option Description Tuning Suggestion
PWX Latency in Select the maximum time, in seconds, that Use the default value.
seconds PowerExchange on the source platform waits
for more change data before flushing data to
PWXPC on the PowerCenter Integration
Service platform.
Default is 2.
Maximum Rows Per Maximum number of change records that To improve efficiency on the PowerCenter
commit PWXPC reads from the source before it Integration Service machine and the target
flushes the data buffer to commit the change databases, reduce commit processing.
data to the targets.
Default is 0, which means that PWXPC does
not use maximum rows.
Minimum Rows Per Minimum number of change records that If your UOWs contain only a few changes, select
commit PowerExchange reads from the change stream a larger value for this option to increase the size
before it passes any commit records to of the UOWs.
PWXPC.
Default is 0, which means that PWXPC does
not use minimum rows.
Offload Processing Select this option to request CDC offload For more information about offload processing,
processing. see “CDC Offload and Multithreaded Processing”
Default is No. on page 159.
Worker Threads If you select Offload Processing, you can also For more information about offload processing,
set this option to have PowerExchange use see “CDC Offload and Multithreaded Processing”
multiple threads for the offloaded processing on page 159.
on the PowerCenter Integration Service
machine. Enter the number of threads that you
want PowerExchange to use.
Default is 0, which means that PowerExchange
does not use multithreaded processing.
Array Size If the Worker Threads value is greater than Use 25.
zero, the size of the storage array, in number Warning: If you specify a large value, have large
of records, for the threads. records, or run many sessions that use
Valid values are from 25 through 100000. multithreaded processing, you might experience
Default is 25. memory shortages on the PowerCenter
Integration Service machine.
TCPIP Activity Timeout Activity timeout. For most applications, use the default of -1,
If no data, other than heartbeat data, is sent or specifying no activity timeout. Instead,
received during this time interval (in seconds), PowerExchange will use heartbeat processing to
PowerExchange aborts the connection and detect failed connections.
indicates a timeout error.
A value of -1 means that no activity timeout is
set.
For more information about connection options, see PowerExchange Interfaces for PowerCenter.
RELATED TOPICS:
¨ “Tuning Commit Processing ” on page 159
¨ “CDC Offload and Multithreaded Processing” on page 159

Tuning Commit Processing
If the PowerCenter session log for a CDC session contains groups of PWXPC flush messages followed by groups
of source-based commit messages from PowerCenter, the CDC session might be reading change data faster than
the data can be processed and written to the targets. To resolve this issue, you can adjust the values that you set
for following commitment control options on the PWX CDC connection:
¨ UOW Count. If the session log contains mostly PWXPC_10081 flush messages, you might need to increase
the value for this option.
¨ Real-time Flush Latency in milli-seconds. If the session log contains mostly PWXPC_10082 flush messages,
you might need to increase the value for this option.
¨ Maximum Rows Per commit. If the session log contains mostly PWXPC_12128 flush messages, you might
need to increase the value for this option.
PWXPC might also flush change data too frequently because the PWX CDC connection in the CDC session uses
too many of the commitment control options. In this case, use a single option to control commit processing and
disable the unused options.
If your change data has many small UOWs, you can use the Minimum Rows Per commit option to create larger
UOWs of more uniform size. PowerExchange and PWXPC can process a few UOWs of larger size more efficiently
than many small UOWs. By using the Minimum Rows Per commit option to increase the size of UOWs, you can
improve CDC processing efficiency.
The following additional factors can also affect the efficiency with which change data is applied to the targets:
¨ Buffer Memory. The DTM Buffer Size and Default Buffer Block Size values can impact the performance of
the CDC session. If you have enabled the collection of performance details in the CDC session, review the
difference between performance counters 4.5 Time in PowerExchange Processing and 4.6 Elapsed Time. If
the elapsed time is much larger that the PowerExchange processing time, buffer memory constraints might
exist.
¨ Target database. The performance of the target database can impact the performance of the CDC session.
Contact your database administrator to ensure that access to the database is optimized.
CDC Offload and Multithreaded Processing

You can use CDC offload processing with the following types of change data extractions:
¨ CDC sessions that use real-time extraction mode
¨ PowerExchange Logger for Linux, UNIX, and Windows
When you use CDC offload processing with real-time extractions, the change data remains on the source system
and PowerExchange moves the column-level processing to the PowerCenter Integration Service machine that
runs the CDC session. For MVS, DB2 for i5/OS, and Oracle sources, PowerExchange also moves the UOW
Cleanser processing to the PowerCenter Integration Service machine.
When you use CDC offload processing with the PowerExchange Logger for Linux, UNIX, and Windows,
PowerExchange does the following processing:
¨ Reads the change data from the source system and stores it in PowerExchange Logger log files
¨ For MVS, DB2 for i5/OS, and Oracle sources, moves the UOW Cleanser processing to the machine on which
the PowerExchange Logger is running
The PowerExchange Logger stores the change data in log files on the Linux, UNIX, or Windows machine. CDC
sessions can then use continuous extraction mode to extract the change data from the PowerExchange Logger log
files instead of from the source system.
CDC Offload and Multithreaded Processing 159

You can use multithreaded processing for CDC sessions that select offload processing. By default,
PowerExchange uses a single thread to process change data on the PowerCenter Integration Service machine.
When you select multithreaded processing, PowerExchange uses multiple threads to process the change records
in each UOW.
Planning for CDC Offload and Multithreaded Processing

Before you configure CDC offload and multithreaded processing, review the following considerations,
requirements, and restrictions.
Restrictions and Requirements for CDC Offload Processing

If you use CDC offload processing, certain restrictions and requirements apply.
Consider the following restrictions and requirements before implementing offload processing:
¨ You must configure CAPI_CONNECTION statements for the data source in the DBMOVER configuration file on
the remote system. For real-time extraction mode, configure the CAPI_CONNECTION statements in the
dbmover.cfg configuration file on the PowerCenter Integration Service machine. For the PowerExchange
Logger for Linux, UNIX, and Windows, configure the CAPI_CONNECTION statements in the dbmover.cfg
configuration file that the PowerExchange Logger uses.
¨ If you set the optional Idle Time attribute on the PWXPC connection, you must specify -1 or 0 as the attribute
value. If you enter a value greater than 0, PWXPC uses 0.
¨ If you use batch extraction mode, Informatica recommends that you set the Idle Time connection attribute to 0
so that the workflow session ends when the end-of-log (EOL) is reached. With this configuration, you can leave
the PowerExchange Logger running continuously.
¨ PowerExchange does not invoke MVS RACF security authorization for change data extraction. Specifically,
PowerExchange does not validate any CAPX.CND profiles for extracting change data when a workflow runs.
However, PowerExchange does validate CAPX.REG profiles during PowerExchange Logger processing.
¨ PowerExchange does not support CDC offload processing for capture registrations that have been created
from data maps that use any of the following options:
- User access methods
- User-defined fields that invoke programs by using the CALLPROG function
- Record-level exits
¨ To capture change data to PowerExchange Logger for Linux, UNIX, and Windows log files, you must configure
capture registrations for partial condense processing. When you define the capture registration in the
PowerExchange Navigator, select Part in the Condense list. If any of the capture registrations for z/OS or i5/
OS data sources specify Full for the Condense option, the PowerExchange Logger ignores them.
¨ For z/OS data sources, if you use offload processing with the PowerExchange Logger for Linux, UNIX, and
Windows and a group definition file, do not include the SCHEMA statement in the group definition file.
PowerExchange does not support SCHEMA statements for z/OS data sources.
¨ Each PowerExchange Logger for Linux, UNIX, and Windows process must read all of the capture registrations
that it uses from a single CCT file on the remote system. Also, each PowerExchange Logger process must
store the names of its log files in a unique CDCT file on the local system.

Considerations for Multithreaded Processing
In specific situations, multithreaded processing might improve performance for a CDC session. Before you
configure multithreaded processing options, review the following considerations:
¨ Use multithreaded processing when the PWX reader thread of a CDC session uses 100% of a single CPU on a
multi-CPU server on the PowerCenter Integration Service platform while processing change data. When a
single CPU is consumed, spreading the PowerExchange processing across multiple threads improves
throughput. Otherwise, additional threads do not improve throughput.
¨ If the network processing between the source and PowerCenter Integration Service machines is slow, try
specifying 1 for the Worker Threads option to help improve throughput. When you specify one or more worker
threads, PowerExchange overlaps network processing with the processing of the change data on the
PowerCenter Integration Service machine.
¨ For optimal performance, the value for the Worker Threads option should not exceed the number of installed
or available processors on the PowerCenter Integration Service machine.
Enabling Offload and Multithreaded Processing for CDC Sessions

To use CDC offload processing and multithreaded processing, you must configure connection options in the CDC
session and CAPI_CONNECTION statements in the PowerExchange DBMOVER configuration file.
To enable CDC offload and multithreaded processing for CDC sessions:
1. Configure the following options on the PWX CDC Real Time application connection for the CDC session:
Connection Option Description
Location Specifies the node name of the system on which the change data resides. This node
name must be the name of a NODE statement in the dbmover.cfg configuration file on the
PowerCenter Integration Service machine.
Offload Processing Specifies whether to use CDC offload processing to move PowerExchange processing for
the change data from the source system to the PowerCenter Integration Service machine.
Select one of the following values:
- No
- Yes
- Auto. PowerExchange determines whether to use offload processing.
Default is No.
Worker Threads When you select CDC offload processing, specifies the number of threads that
PowerExchange uses on the PowerCenter Integration Service machine to process change
data. You must also enter a value for the Array Size.
Default is 0.
Array Size If the Worker Threads value is greater than zero, specifies the size of the storage array
for each thread, in numbers of records.
Default is 25.
CAPI Connection Name Specifies the name of the source CAPI_CONNECTION statement in the dbmover.cfg on
the PowerCenter Integration Service machine.
2. Copy the CAPI_CONNECTION statements from the DBMOVER configuration file on the source system to the
dbmover.cfg configuration file on the PowerCenter Integration Service machine. For MVS sources, remove all
MVS-specific parameters from the UOWC CAPI_CONNECTION statement.

Use the following table to select the correct CAPI_CONNECTION statement types to configure, based on
source type:
CDC Source Type CAPI_CONNECTION Statements
DB2 for i5/OS AS4J and UOWC
DB2 for Linux, UNIX, and Windows UDB
Microsoft SQL Server MSQL
MVS sources LRAP and UOWC
Oracle ORCL and UOWC
Configuring PowerExchange to Capture Change Data on a Remote

System
You can use CDC offload processing with the PowerExchange Logger for Linux, UNIX, and Windows to capture
change data from source systems other than the system where the PowerExchange Logger runs. With CDC
offload processing, a PowerExchange Logger for Linux, UNIX, and Windows can capture change data from i5/OS
and MVS systems as well as from other Linux, UNIX, or Windows systems.
CDC sessions use continuous extraction mode to extract the change data from the PowerExchange Logger log
files instead of from the source system.
You must first install PowerExchange on the remote Linux, UNIX, or Windows system.
Before you start a PowerExchange Logger for Linux, UNIX, and Windows process on a remote system, configure
the pwxccl.cfg and the dbmover.cfg configuration files on that system. When you use CDC offload processing,
each PowerExchange Logger must have unique pwxccl.cfg and dbmover.cfg configuration files.
To extract the change data from the PowerExchange Logger on the remote system, you must also configure and
start a PowerExchange Listener on that system. The dbmover.cfg file that the PowerExchange Listener uses must
specify the same CAPT_PATH value as the dbmover.cfg file that the PowerExchange Logger uses. Alternatively,
you can use the same dbmover.cfg file for the PowerExchange Logger and the PowerExchange Listener.
The following steps describe how to configure a PowerExchange Logger and PowerExchange Listener to offload
change data from source systems and capture that data to PowerExchange Logger log files on Linux, UNIX, or
Windows.
RELATED TOPICS:
¨ “Extracting Change Data Captured on a Remote System” on page 168
Configuring pwxccl.cfg
Configure the pwxccl.cfg configuration file for the PowerExchange Logger on the remote system where the
PowerExchange Logger will run.
PowerExchange provides a sample pwxccl.cfg file in the PowerExchange installation directory, which you can
copy and then edit. For CDC offload processing, customize the following parameters:

CAPTURE_NODE
Specifies the node name of the system on which the change data was originally captured.
This node name must match the node name in a NODE statement in the dbmover.cfg configuration file that
the PowerExchange Logger uses.
CAPTURE_NODE_EPWD
Specifies an encrypted password for the CAPTURE_NODE_UID user ID.
If you specify CAPTURE_NODE_UID, you must specify a password for that user ID by using either
CAPTURE_NODE_EPWD or CAPTURE_NODE_PWD. If you specify CAPTURE_NODE_EPWD, do not also
specify CAPTURE_NODE_PWD.
Tip: You can create an encrypted password in the PowerExchange Navigator by selecting File > Encrypt
Password.
CAPTURE_NODE_PWD
Specifies a clear text password for the CAPTURE_NODE_UID user ID.
If you specify CAPTURE_NODE_UID, you must specify a password for that user ID by using either
CAPTURE_NODE_EPWD or CAPTURE_NODE_PWD. If you specify CAPTURE_NODE_PWD, do not also
specify CAPTURE_NODE_EPWD.
CAPTURE_NODE_UID
Specifies a user ID that permits PowerExchange to read capture registrations and change data on the remote
node that is specified in the CAPTURE_NODE parameter. Whether this parameter is required depends on the
operating system of the remote node and the SECURITY setting in the DBMOVER configuration file for the
PowerExchange Listener on that node.
If the CAPTURE_NODE is an MVS or i5/OS system with a SECURITY setting of 1 or 2, you must specify a
valid operating system user ID. If the SECURITY setting is 2, PowerExchange uses the specified user ID to
control access to capture registrations and change data. However, if the SECURITY setting is 1,
PowerExchange uses the user ID under which the PowerExchange Listener job runs.
If the CAPTURE_NODE is an MVS or i5/OS system with a SECURITY setting of 0, do not specify this
parameter. PowerExchange uses the user ID under which the PowerExchange Listener job runs to control
access to capture registrations and change data.
If the CAPTURE_NODE is a Linux, UNIX, or Windows system, specify a user ID that is valid for the data
source type:
¨ For a DB2 for Linux, UNIX, or Windows source, enter a valid operating system user ID that has DB2
DBADM or SYSADM authority.
¨ For an Oracle source, enter a database user ID that permits access to Oracle redo logs and Oracle
LogMiner.
¨ For a SQL Server instance that uses SQL Server Authentication, enter a database user ID that permits
access to the SQL Server distribution database. For a SQL Server instance that uses Windows
Authentication, PowerExchange uses the user ID under which the PowerExchange Listener was started. In
this case, do not specify this parameter unless you want to specify another user.
CHKPT_BASENAME
Specifies an existing path and base name file name to use for generating the PowerExchange Logger
checkpoint files.
CONDENSENAME
Optional. Specifies a name for the command-handling service for a PowerExchange Logger for Linux, UNIX,
and Windows process to which you issue pwxcmd commands.

This service name must match the service name in the associated SVCNODE statement in the DBMOVER
configuration file.
CONN_OVR
Specifies the name of the CAPI_CONNECTION statement in the dbmover.cfg file that the PowerExchange
Logger uses. This CAPI_CONNECTION statement defines the connection to the change stream for the data
source type.
For data sources that include UOW Cleanser (UOWC) CAPI_CONNECTION statements, specify the name of
this statement. For all other data sources, specify the CAPI_CONNECTION name for the data source type.
DB_TYPE
Specifies the data source type.
Use the following table to select the correct DB_TYPE to configure, based on source type:
CDC Source Type DB_TYPE Value
Adabas ADA
Datacom DCM
DB2 for i5/OS AS4
DB2 for z/OS DB2
IDMS log-based IDL
IMS IMS
Microsoft SQL Server MSS
Oracle ORA
VSAM VSM
DBID
Specifies the source collection identifier that is defined in the registration group. The PowerExchange
Navigator displays this value in the Resource Inspector when you open the registration group. When used
with DB_TYPE, it defines selection criteria for capture registrations in the CCT file.

Use the following table to select the correct DBID value, based on source type:
CDC Source Type DBID Value
Adabas The Instance name that is displayed for the registration group in the
Datacom One of the following values:

- The MUF Name value that is displayed for the registration group in
the PowerExchange Navigator.
- For Datacom synchronous CDC, the MUF parameter value in the
DTLINPUT data set specified in the MUF JCL.
- For Datacom table-based CDC, the REG_MUF parameter value in the
ECCRDCMP member of the RUNLIB library.
DB2 for i5/OS One of the following values:

- The Instance name that is displayed for the registration group in the
- The INST parameter value in the AS4J CAPI_CONNECTION
statement in the DBMOVER member of the CFG file.
DB2 for Linux, UNIX, and Windows The Database name that is displayed for the registration group in the
DB2 for z/OS One of the following values:

- The Instance name that is displayed for the registration group in the
- The RN parameter value from the DB2 statement in the REPDB2OP
member of the RUNLIB library.
IDMS Log-based One of the following values:

- The Logsid value that is displayed for the registration group in the
- The LOGSID parameter value in the ECCRIDLP member of the
RUNLIB library.
IMS One of the following values:

- The IMSID value that is displayed for the registration group in the
- For IMS log-based CDC, the first parameter of the IMSID statement in
the CAPTIMS member of the RUNLIB library.
Microsoft SQL Server The Instance name that is displayed for the registration group in the
VSAM The Instance name that is displayed for the registration group in the
EPWD
A deprecated parameter. Use CAPTURE_NODE_EPWD instead. If both CAPTURE_NODE_EPWD and

EPWD are specified, CAPTURE_NODE_EPWD takes precedence.
EXT_CAPT_MASK
Specifies an existing path and unique prefix to be used for generating the PowerExchange Logger log files.

PWD
A deprecated parameter. Use CAPTURE_NODE_PWD instead. If both CAPTURE_NODE_PWD and PWD are
specified, CAPTURE_NODE_PWD takes precedence.
RESTART_TOKEN and SEQUENCE_TOKEN
Optional. Specifies a restart point for starting change data processing when the PowerExchange Logger is
cold started.
The format of the restart tokens varies based on data source type and, if specified, must match the format
required by the DB_TYPE specified. If you do not specify these parameters, the PowerExchange Logger uses
the end of the change stream as the restart point when cold started.
UID
A deprecated parameter. Use CAPTURE_NODE_UID instead. If both CAPTURE_NODE_UID and UID are
specified, CAPTURE_NODE_UID takes precedence.
RELATED TOPICS:
Configuring dbmover.cfg on the PowerExchange Logger Machine

On the remote system where the PowerExchange Logger will run, configure the dbmover.cfg file that the
PowerExchange Logger and PowerExchange Listener will use.
Note: Unless the change data is captured on the PowerCenter Integration Service machine, you must run a
PowerExchange Listener so CDC sessions can extract the offloaded change data.
The dbmover.cfg file that the PowerExchange Listener uses must specify the same CAPT_PATH value as the
dbmover.cfg that the PowerExchange Logger uses. Alternatively, you can use the same dbmover.cfg configuration
file for the PowerExchange Logger and PowerExchange Listener. This step assumes that you use the same
dbmover.cfg file.
PowerExchange provides a sample dbmover.cfg file in the PowerExchange installation directory, which you can
copy and then edit. For CDC offload processing, set the following parameters:
CAPT_PATH
Specifies the path to the directory where the CDCT file resides. The CDCT file contains information about the
PowerExchange Logger log files, such as file names and number of records.
Each PowerExchange Logger that uses CDC offload processing to capture change data requires its own
CDCT file.
CAPX CAPI_CONNECTION
Specifies parameters for continuous extraction of change data from PowerExchange Logger log files. In
continuous extraction mode, extractions run in near real time and read the data in the PowerExchange Logger
log files as the change stream.
In the DFLTINST parameter of the CAPX CAPI_CONNECTION, specify the DBID value from the
PowerExchange Logger pwxccl.cfg configuration file.
LOGPATH
Specifies the path to the PowerExchange log files that contain PowerExchange Logger messages.
NODE
Specifies the TCP/IP connection information for a PowerExchange Listener.

Configure a NODE statement for the system on which the change data was originally captured. Specify the
node name for this statement in the CAPTURE_NODE parameter of the PowerExchange Logger pwxccl.cfg
configuration file.
Source-specific CAPI_CONNECTION
Specifies CAPI parameters that are specific to the data source type and that PowerExchange uses to connect
to the change stream.
Copy the CAPI_CONNECTION statements from the DBMOVER configuration file on the source system where
the change data resides. Use the following table to select the correct CAPI_CONNECTION statement types to
configure, based on source type:
CDC Source Type CAPI_CONNECTION Statements
DB2 for i5/OS AS4J and UOWC
Microsoft SQL Server MSQL
MVS sources LRAP and UOWC
For MVS sources, remove MVS-specific parameters from the UOWC CAPI_CONNECTION statement.
SVCNODE
Optional. Specifies the TCP/IP port on which a command-handling service for a PowerExchange Listener or
PowerExchange Logger for Linux, UNIX, and Windows process listens for pwxcmd commands.
TRACING
Optional. Enables alternative logging. By using alternative logging, you can separate PowerExchange Logger
messages from other PowerExchange messages.
Configuring dbmover.cfg on the PowerCenter Integration Service Machine

In the dbmover.cfg configuration file on the PowerCenter Integration Service machine, add a NODE statement for
the PowerExchange Listeners that run on the following systems:
¨ The system where the change data was originally captured and where the capture registrations reside
¨ The system where the change data is stored in PowerExchange Logger for Linux, UNIX, and Windows log files
Configuring Capture Registrations for the PowerExchange Logger

For the PowerExchange Logger on Linux, UNIX, and Windows to capture change data from a remote system,
capture registrations for the remote source must specify Part for the Condense option.
If capture registrations do not specify Part for the Condense option, delete the capture registrations and
corresponding extraction maps. Then create the capture registrations again. PowerExchange generates
corresponding extraction maps. You can edit the PowerExchange-generated extraction maps or create additional
ones.
Tip: Do not add DTL_BI or DTL_CI columns to the extraction maps if you set the CAPT_IMAGE parameter to AI in
the pwxccl.cfg configuration file. With the AI setting, the PowerExchange Logger captures after images only.

Consequently, PowerExchange cannot populate BI columns with before images. Also, with this setting,
PowerExchange writes Nulls to CI columns for any INSERT or DELETE operations.
Starting the PowerExchange Logger and PowerExchange Listener

Start the PowerExchange Logger and PowerExchange Listener on the remote system that will capture the change
data.
Note: If the remote system also runs the PowerCenter Integration Service, you can use local mode to extract the
data instead of a PowerExchange Listener.
Extracting Change Data Captured on a Remote System

After you have captured change data on a remote system in the PowerExchange Logger for Linux, UNIX, and
Windows log files, you can use continuous extraction mode to extract the change data in a CDC session. In the
CDC session, select the appropriate PWX CDC Real Time connection for the source type. For example, if you
captured change data for a DB2 for Linux, UNIX, and Windows source to PowerExchange Logger log files on a
remote system, use a PWX DB2LUW CDC Real Time connection to extract the data.
Customize the following connection options to extract offloaded change data:
¨ Location. Specify the node name for the PowerExchange Listener that runs on the remote system where the
change data was stored in PowerExchange Logger log files.
¨ Map Location. Specify the node name for the PowerExchange Listener that runs on the source system where
the change data was originally captured. The PowerExchange Listener on the original source system stores the
capture registrations.
¨ Map Location User and Map Location Password. Specify a user ID and password that can access the
capture registrations for the change data.
If the PowerExchange Listener on the source system is running on MVS or i5/OS and is configured with
security, specify a valid operating system user ID. You do not need to specify this parameter if the
PowerExchange Listener is running without security.
If the PowerExchange Listener on the data source system is running on Linux, UNIX, or Windows, specify a
valid database user ID.
¨ CAPI Connection Name Override. Specify the name of the CAPX CAPI_CONNECTION in the dbmover.cfg
configuration file used by the PowerExchange Listener on the remote system where the change data is stored
in PowerExchange Logger log files.
For more information about configuring PWX CDC Real Time application connections, see PowerExchange
Interfaces for PowerCenter.
Configuration File Examples for CDC Offload Processing

The following examples show the configuration required for CDC offload processing.
Extracting Change Data from Oracle Using CDC Offload Processing - Example
In this example, a CDC session that uses real-time connections to extract change data from an Oracle source is
changed to use CDC offload processing. The source change data remains on Oracle system but all column-level
and UOW Cleanser processing is moved to the PowerCenter Integration Service machine.
The Oracle system has the following CAPI_CONNECTION statements in the dbmover.cfg configuration file that
the PowerExchange Listener uses to read change data:
/* UOW Cleanser
CAPI_CONNECTION=(NAME=UOWCORA,TYPE=(UOWC,CAPINAME=CAPIORA,RSTRADV=600))

/* Oracle CDC
CAPI_CONNECTION=(NAME=CAPIORA,TYPE=(ORCL,catint=120,ORACOLL=PRODORA))
To extract change data from Oracle using CDC offload processing:
1. Configure the dbmover.cfg configuration file on the PowerCenter Integration Service machine for CDC offload
processing.
Copy the UOWC and ORCL CAPI_CONNECTION statements from the dbmover.cfg file on the Oracle system
to the dbmover.cfg configuration file on the PowerCenter Integration Service machine. In this example, the
following CAPI_CONNECTION statements are copied into the dbmover.cfg:
2. Stop the CDC session.
3. Update the following options on the PWX CDC Real Time application connection in the CDC session:
¨ Select Yes for the Offload Processing option.
¨ In the CAPI Connection Name option, specify the name of the UOWC CAPI_CONNECTION statement. In
this example, the name is UOWCORA.
4. Restart the CDC session.
Capturing and Extracting Change Data from a Remote UNIX System - Example
In this example, change data for Oracle sources is captured by the PowerExchange Logger for Linux, UNIX, and
Windows on a different UNIX system from where the Oracle instance runs. The Oracle sources are registered for
capture on the UNIX system where the Oracle instance runs. A CDC session then extracts the change data for the
Oracle sources from PowerExchange Logger log files on the remote UNIX system, rather than from the system
where the change data was originally captured.
The original UNIX system has the following CAPI_CONNECTION statements in the dbmover.cfg file that the
PowerExchange Listener uses to read change data:
/* UOW Cleanser
/* Oracle CDC
The instance name used to register the Oracle tables for capture on the original UNIX system is called PRODORA.
The following procedure assumes that PowerExchange is installed and configured on the remote UNIX system
where the PowerExchange Logger for Linux, UNIX, and Windows will run.
To capture and extract change data from a remote UNIX system:
1. Configure the PowerExchange Logger for Linux, UNIX, and Windows on the remote UNIX system by
completing the following steps:
¨ Configure pwxccl.cfg.
¨ Configure dbmover.cfg on the PowerExchange Logger machine.
In this example, the dbmover.cfg on the remote UNIX system has the following parameters:
/*
/* dbmover.cfg
/*
LISTENER=(unix1,TCPIP,2480)
NODE=(ORA2,TCPIP,prodora2,2480)
...
logpath=/pwx/logs/oracond
CAPT_XTRA=/pwx/capture/oracond/camaps
CAPT_PATH=/pwx/capture/oracond
ORACLEID=(PRODORA,ORAINST2,ORAINST2,ORAINST2)
/*
/* Source-specific CAPI Connection

/*
/* CAPX CAPI Connection for continuous extraction
CAPI_CONNECTION=(NAME=CAPXORA,TYPE=(CAPX,DFLTINST=PRODORA,FILEWAIT=60,RSTRADV=600))
In this example, the pwxccl.cfg file has the following parameters:
/*
/* pwxccl.cfg
/*
DBID=PRODORA
DB_TYPE=ORA
CONN_OVR=UOWCORA
CAPTURE_NODE=ORA2
CAPTURE_NODE_UID=orauser
CAPTURE_NODE_PWD=orapwd
EXT_CAPT_MASK=/pwx/capture/oracond/condense
CHKPT_NUM=3
CHKPT_BASENAME=/pwx/capture/oracond/condense.chkpt
COND_CDCT_RET_P=50
COLL_END_LOG=0
NO_DATA_WAIT=1
NO_DATA_WAIT2=2
FILE_SWITCH_VAL=200000
FILE_SWITCH_CRIT=R
CAPT_IMAGE=BA
SIGNALLING=N
UID=orauser
PWD=orapwd
VERBOSE=Y
2. After you configure the dbmover.cfg and the pwxccl.cfg configuration files, start the PowerExchange Listener
and PowerExchange Logger on the remote UNIX system.
3. On the PowerCenter Integration Service machine, customize the following statements:
¨ NODE statement to point to the PowerExchange Listener on the remote UNIX system, which is where the
PowerExchange Logger runs.
¨ NODE statement to point to the PowerExchange Listener on the original UNIX system, which is where the
Oracle instance runs and the tables are registered for capture.
In this example, the following statements are added to the dbmover.cfg on the PowerCenter Integration
Service machine:
NODE=(unix1,TCPIP,unix1,2480)
NODE=(ORA2,TCPIP,prodora2,2480)
4. Create and configure the PowerCenter mapping, session, and workflow to extract the change data.
5. To extract the change data from the remote UNIX system, configure a PWX Oracle CDC Real Time
application connection in the CDC session.
In this example, specify the following options to point to the remote UNIX system for the change data, the
original UNIX system for the extraction maps, and the CAPX CAPI_CONNECTION name to use continuous
extraction mode:
¨ For the Location option, specify unix1.
¨ For the Map Location option, specify ORA2.
¨ For the Map Location User option, specify a valid Oracle user ID.
¨ For the Map Location Password option, specify the password for the Oracle user ID.
¨ For the CAPI Connection Name option, specify CAPXORA.
Cold start the CDC session to extract the change data from the PowerExchange Logger log files on the
remote UNIX system.

INDEX
A
recovery example 146
restart points for warm starts 115
alternative logging 25, 43 restart token file 137
application name stopping 142
configuring for CDC sessions 132 tuning 154
application names 113 CDCT file 21, 53, 54
architectural diagrams change data capture (CDC)
batch or continuous extraction processing 8 architecture 8
real-time extraction processing 8 data source types 4
architecture, PowerExchange CDC 8 DB2 for Linux, UNIX, and Windows CDC 56
archive log destination 84 Oracle LogMiner CDC 80
ARCHIVELOG mode overview 2
enabling for Oracle LogMiner CDC 84 PowerExchange components 6
SQL Server CDC 70
task summary 10
B
change data extraction
creating restart tokens for extractions 135
batch extraction mode 106 extracting data captured from a remote system 168
extraction modes 106
monitoring in PowerCenter 151
C
monitoring in PowerExchange 148
offload processing 159
cache files 23 overview 3
CAPI connection statements overview of extracting change data 125
CAPI_CONNECTION statement 12 task flow 126
CAPI_SRC_DFLT statement 12 testing extraction maps 126
CAPX parameters 14 tuning CDC sessions 154
introduction 14 checkpoint files 22, 53
MEMCACHE parameter 155 close (pwxcmd) 16
MSQL CAPI_CONNECTION statement 76 closeforce (pwxcmd) 16
ORCL CAPI_CONNECTION statement 91, 92 commit processing
RSTRADV parameter 155 configuring for CDC sessions 133
UDB CAPI_CONNECTION statement 62 controlling with connection attributes 119
UOWC parameters 99 examples 121
capture catalog table in CDC sessions 118
creating 60 minimum and maximum rows per commit 120
DTLUCUDB SNAPSHOT command 61 target latency 121
initializing the table 61 tuning 159
capture registrations compatible parameter 84
grouping in PowerExchange Logger group definition file 44 components, PowerExchange
settings for the PowerExchange Logger 27 for CDC 6
CAPX CAPI_CONNECTION parameters PowerExchange Listener 6, 12
parameters and syntax 14 PowerExchange Logger 6
catalog, Oracle PowerExchange Navigator 7
copying for Oracle LogMiner CDC 87 configuration tasks
parameters in ORCL CAPI_CONNECTION 97 DB2 for Linux, UNIX, and Windows CDC 58, 59
CDC data map Oracle LogMiner CDC 83
extraction map 137 PowerExchange Listener 12
CDC sessions PowerExchange Logger 27
buffer memory 159 SQL Server CDC 73
commit processing 118 continuous extraction mode 106
default restart points 114 Controller task, PowerExchange Logger 20
methods of starting 113, 140
monitoring in PowerCenter 151
monitoring in PowerExchange 148
offload processing 123, 159
171
D E
data maps extraction map columns, PowerExchange-generated
use in DB2 for Linux, UNIX, and Windows CDC 65 DTL__BI_columnname 106
data sources, types 4 DTL__CAPXACTION 106
database row tests 126 DTL__CAPXCASDELIND 106
datatypes DTL__CAPXRESTART1 106
SQL Server 71 DTL__CAPXRESTART2 106
DB2 for Linux, UNIX, and Windows CDC DTL__CAPXRRN 106
changing a source table definition 66 DTL__CAPXTIMESTAMP 106
configuring in DB2 58 DTL__CAPXUOW 106
configuring in PowerExchange with the Logger 60 DTL__CAPXUSER 106
configuring in PowerExchange without the Logger 59 DTL__CI_columnname 106
creating the capture catalog table 60 extraction maps
dbmover.cfg parameters 61 PowerExchange-generated columns 106
example dbmover.cfg statements 62 extraction modes 106
IBM APARs 69 extraction of change data
initializing the capture catalog table 61 creating restart tokens for extractions 135
overview 56 extracting data captured from a remote system 168
planning 57 extraction modes 106
prerequisites 57 monitoring in PowerCenter 151
restrictions 58 monitoring in PowerExchange 148
stopping 66 offload processing 159
troubleshooting 69 overview of extracting change data 125
user authority requirement 57 task flow 126
using a data map 65 testing extraction maps 126
DB2 partitioned databases tuning CDC sessions 154
reconfiguring 67
DB2 SQL1224 error 69
DB2CODEPAGE environment variable 58
DB2NOEXITLIST environment variable 58 F
dbmover.cfg file switches
APPBUFSIZE 155 description 25
CAPI_CONNECTION statements 12 FILESWITCH command 49
CAPI_SRC_DFLT statement 12
CAPT_PATH parameter 43
CAPT_PATH statement 12
CAPT_XTRA statement 12 G
COMPRESS parameter 155 group definition file
DB2 for Linux, UNIX, and Windows CDC parameters 61 configuring for PowerExchange Logger 44
DB2 for Linux, UNIX, and Windows example statements 62 example file 46
general CDC parameters 12 GROUP statement 45
LOGPATH parameter 43 REG statement 45
Oracle LogMiner CDC example statements 90 SCHEMA statement 45
Oracle LogMiner CDC parameters 90 statements and parameters 45
PowerExchange Logger parameters 43 group source
SQL Server CDC example statements 76 description 116
SQL Server CDC parameters 75 processing CDC data for multiple source definitions 117
SVCNODE parameter 43
TRACE parameter 155
TRACING parameter 43
types of CAPI connection statements for CDC 14 I
detail.log 25 idle time
diagrams configuring for a CDC session 131
batch or continuous extraction processing 8 description 131
real-time extraction processing 8 integration with PowerCenter 7
DISPLAY ACTIVE command 150
DTL__CAPXRESTART1
sequence token 135
DTL__CAPXRESTART2 L
restart token 135 listtask (pwxcmd) 17, 150
DTLUAPPL lock files 23
displaying restart tokens 135 log files of PowerExchange Logger
DTLUCUDB SNAPSHOT command 61 file switches 25
DTLUTSK utility 142 log files, PowerExchange Logger
maintaining 53
naming 22
LogMiner, Oracle
172 Index
configuring for Oracle CDC 86 enabling ARCHIVELOG mode 84
example dbmover.cfg statements 90
overview 80
M performance considerations 83
planning 81
maximum row count restrictions and requirements 81
configuring for a CDC session 133 SQL*Loader restrictions 82
message log files 25 stopping 102
Microsoft SQL Server CDC supplemental logging requirement 86
changing a source table definition 79 supported datatypes 81
configuration tasks 73 transaction_auditing parameter 84
configuring in PowerExchange with the Logger 75 user privileges required 85
configuring in PowerExchange without the Logger 74 ORCL CAPI_CONNECTION statement
datatypes supported 71 CATBEGIN parameter 97
dbmover.cfg parameters 75 CATEND parameter 97
example dbmover.cfg statements 76 CATINT parameter 97
overview 70 Oracle catalog parameters 97
planning 71 parameters and syntax 91, 92
prerequisites 71 output files, PowerExchange Logger
restrictions 73 cache files 23
stopping 78 CDCT file 21
user authority requirements 71 checkpoint files 22
minimal global supplemental logging 86
minimum row count
configuring for a CDC session 133
monitoring CDC sessions P
PowerCenter output to monitor 151 partitioned DB2 database
PowerCenter session log messages 151 reconfiguring 67
PowerExchange extraction statistics messages 149 performance
PowerExchange multithreaded processing statistics 149 CDC session performance details 151
PowerExchange output to monitor 148 offload processing and multithreaded processing 159
PowerExchange read progress messages 148 Oracle LogMiner CDC considerations 83
viewing performance details in PowerCenter 151 PowerExchange Client for PowerCenter (PWXPC) 7
MSQL CAPI_CONNECTION statement PowerExchange components
parameters and syntax 76 for CDC 6
multithreaded processing PowerExchange Listener 6, 12
enabling for CDC sessions 161 PowerExchange Logger 6
overview 123, 159 PowerExchange Navigator 7
planning considerations 160 PowerExchange Listener
restrictions and requirements 160 CLOSE command 16
statistics messages 149 DISPLAY ACTIVE command 17, 150
displaying active listener tasks 17
overview 12
O starting 16
stopping 16
offload processing STOPTASK command 16
configuration examples 168 PowerExchange Logger for Linux, UNIX, and Windows
enabling for CDC sessions 161 assessing performance 52
Logger capture of changes from a remote source 162 backing up CDCT, checkpoint, and log files 54
overview 123, 159 batch mode 26
planning considerations 160 cache files 23
restrictions and requirements 160 CDCT file 21
oracapt_rac.sql 83 change capture from a remote source 162
oracapt.sql 83 checkpoint files 22
Oracle CDC cold starting 49
configuring Oracle LogMiner 86 CONDENSE command 49
Oracle LogMiner configuring 27
configuring for Oracle CDC 86 continuous mode 26
Oracle LogMiner CDC controlling 49
archive log destination 84 dbmover.cfg parameters 43
changing a source table definition 102 DISPLAY ALL command 49
compatible parameter 84 DISPLAY CHECKPOINTS command 49
configuration in a RAC environment 87 DISPLAY CPU command 49
configuration script files 83 DISPLAY EVENTS command 49
configuring in Oracle 83 DISPLAY MEMORY command 49
configuring PowerExchange with the Logger 89 DISPLAY RECORDS command 49
configuring PowerExchange without the Logger 88 DISPLAY STATUS command 49
copying the Oracle catalog 87 extracting remotely captured changes from Logger log files 168
dbmover.cfg parameters 90
Index 173
FILESWITCH command 49 UID parameter 29
group definition file 44 VERBOSE parameter 29
lock files 23 pwxcmd
log file switches 25 close 16
log files 22 closeforce 16
maintaining CDCT file and log files 53 listtask 17
memory requirement on Linux and UNIX 27 listtask command 150
message log files 25 PWXPC 7
offload processing 162
operational modes 25
output files 21
overview 19 R
pwxccl.cfg parameters 28 real application clusters (RACs)
regenerating the CDCT file after a failure 54 configuring for Oracle LogMiner CDC 87
required capture registration settings 27 real-time extraction mode 106
running in background mode on Linux or UNIX 27 real-time flush latency
SHUTCOND command 49 configuring for a CDC session 133
SHUTDOWN command 49 reconfiguring DB2 partitioned database 67
start point in change stream 48 recovery
starting 47 example 146
stopping 49 PM_REC_STATE table 111, 112
subtasks 20 PM_RECOVERY table 111
PowerExchange-generated extraction map columns PM_TGT_RUN_ID table 111
DTL__BI_columnname 106 recovery information for nonrelational targets 112
DTL__CAPXACTION 106 recovery state file for nonrelational targets 113
DTL__CAPXCASDELIND 106 recovery tables for relational targets 111
DTL__CAPXRESTART1 106 restart
DTL__CAPXRESTART2 106 $PMRootDir/Restart 132, 136
DTL__CAPXTIMESTAMP 106 application name 132
DTL__CAPXUOW 106 default restart points 114
DTL__CAPXUSER 106 earliest restart points 114
DTL__CI_columnname 106 methods of starting CDC sessions 113, 140
DTL__columnname_CNT 106 null restart tokens 114
DTL__columnname_IND 106 restart token file 110, 132
pwxccl statement restart token file folder 132
parameters 48 RESTART1 138
syntax 47 RESTART2 138
pwxccl.cfg restart points
CAPT_IMAGE parameter 29 defaults 114
CAPTURE_NODE parameter 29 earliest 114
CAPTURE_NODE_EPWD parameter 29 restart token
CAPTURE_NODE_PWD parameter 29 DTL__CAPXRESTART2 135
CAPTURE_NODE_UID parameter 29 restart token file
CHKPT_BASENAME parameter 29 example 139
CHKPT_NUM parameter 29 explicit override 137
COLL_END_LOG parameter 29 overview 109
COND_CDCT_RET_P parameter 29 special override 138
CONDENSE_SHUTDOWN_TIMEOUT parameter 29 syntax 137
CONDENSENAME parameter 29 restart tokens
configuring 28 creating for extractions 135
CONN_OVR parameter 29 displaying with DTLUAPPL 135
DB_TYPE parameter 29 DTL__CAPXRESTART1 135
DBID parameter 29 DTL__CAPXRESTART2 135
example file 42 null 114
EXT_CAPT_MASK parameter 29 overview 109
FILE_FLUSH_VAL parameter 29 recovery state file 113
FILE_SWITCH_CRIT parameter 29 recovery state table 112
FILE_SWITCH_MIN parameter 29 row tests 126
FILE_SWITCH_VAL parameter 29
GROUPDEFS parameter 29
LOGGER_DELETES_EXPIRED_CDCT_RECORDS parameter 29
MAX_RETENTION_EXPIRY_DAYS parameter 29 S
NO_DATA_WAIT parameter 29 sequence token
NO_DATA_WAIT2 parameter 29 DTL__CAPXRESTART1 135
parameters 28 SHOW_THREAD_PERF parameter 149
PROMPT parameter 29 source RDBMSs 4
RESTART_TOKEN parameter 29 source table definitions
SEQUENCE_TOKEN parameter 29 changing a DB2 table definition 66
SIGNALLING parameter 29
174 Index
changing a SQL Server table definition 79 DB2 for Linux, UNIX, and Windows CDC 69
changing an Oracle table definition 102 tuning CDC sessions
SQL Server CDC APPBUFSIZE parameter 155
changing a source table definition 79 buffer memory 159
configuration tasks 73 CAPI_CONNECTION MEMCACHE parameter 155
configuring in PowerExchange with the Logger 75 CAPI_CONNECTION RSTRADV parameter 155
configuring in PowerExchange without the Logger 74 commit processing tuning 159
datatypes supported 71 COMPRESS parameter 155
dbmover.cfg parameters 75 DBMOVER tuning parameters 155
example dbmover.cfg statements 76 methods 154
overview 70 PWX CDC connection options 157
planning 71 TRACE parameter 155
prerequisites 71
restrictions 73
stopping 78
user authority requirements 71 U
SQL*Loader UDB CAPI_CONNECTION statement
restrictions for Oracle CDC 82 parameters and syntax 62
STOPTASK command UOW count
CDC sessions, stopping 142 configuring for a CDC session 133
supplemental logging, Oracle 86 UOWC CAPI_CONNECTION parameters
parameters and syntax 99
user authority
T DB2 for Linux, UNIX, and Windows CDC requirement 57

Oracle LogMiner CDC requirements 85
task flow SQL Server CDC requirement 71
CDC implementation 10
extracting change data 126
terminating conditions
idle time for CDC sessions 131 W
testing a change data extraction 126 warm starts
transaction_auditing parameter 84 CDC session restart points 115
troubleshooting
Index 175

CDC Guide For Linux, Unix, and Windows: Informatica Powerexchange (Version 9.0)

Uploaded by

Copyright:

Available Formats

CDC Guide For Linux, Unix, and Windows: Informatica Powerexchange (Version 9.0)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

CDC Guide For Linux, Unix, and Windows: Informatica Powerexchange (Version 9.0)

Uploaded by

Copyright:

Available Formats

Informatica PowerExchange (Version 9.

CDC Guide for Linux, UNIX, and Windows

Copyright (c) 1998-2009 Informatica. All rights reserved.

Part Number: PWX-CCl-900-0001

Part I: PowerExchange CDC Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Chapter 1: Change Data Capture Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

Part II: PowerExchange CDC Components. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Chapter 2: PowerExchange Listener. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Part III: PowerExchange CDC Data Sources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Chapter 5: Microsoft SQL Server Change Data Capture. . . . . . . . . . . . . . . . . . . . . . 70

Chapter 6: Oracle Change Data Capture with Oracle LogMiner. . . . . . . . . . . . . . . . 80

Table of Contents iii

Part IV: Change Data Extraction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

Chapter 7: Introduction to Change Data Extraction. . . . . . . . . . . . . . . . . . . . . . . . . 105

Chapter 8: Extracting Change Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

Chapter 9: Managing Change Data Extractions. . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

Chapter 10: Monitoring and Tuning Options. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148

¨ PowerExchange for DB2 ® for Linux®, UNIX®, and Windows®

¨ PowerExchange for Oracle ®

¨ PowerExchange for SQL Server ®

Informatica Customer Portal

Informatica Web Site

Informatica How-To Library

Informatica Knowledge Base

Informatica Multimedia Knowledge Base

Informatica Global Customer Support

Toll Free Toll Free Toll Free

Standard Rate Standard Rate Standard Rate

¨ Change Data Capture Introduction, 2

Change Data Capture Introduction

¨ PowerExchange CDC Overview, 2

¨ PowerExchange CDC Components, 6

¨ PowerExchange Integration with PowerCenter, 7

¨ PowerExchange CDC Architecture, 8

¨ Summary of CDC Implementation Tasks, 10

PowerExchange CDC Overview

These sources are:

¨ DB2 for Linux, UNIX, and Windows

¨ Microsoft SQL Server on Windows

¨ Oracle on Linux, UNIX, or Windows

The change data replication process consists of following high-level steps:

Change Data Extraction and Apply

PowerExchange CDC Overview 3

Extraction Mode Description

PowerExchange CDC Data Sources

DB2 for Linux, UNIX, and Windows Data Sources

4 Chapter 1: Change Data Capture Introduction

Microsoft SQL Server Data Sources

Oracle Data Sources

i5/OS and z/OS Data Sources with Offload Processing

PowerExchange CDC Data Sources 5

¨ PowerExchange Navigator. Required.

¨ CCT file for capture registrations

¨ CAMAPS directory for extraction maps

¨ DATAMAPS directory for DB2 data maps

PowerExchange Logger for Linux, UNIX, and Windows

6 Chapter 1: Change Data Capture Introduction

PowerExchange Integration with PowerCenter

Note: This guide assumes that you use PWXPC.