ISSN: 2393-8528

Contents lists available at www.ijicse.in

International Journal of Innovative Computer Science & Engineering

Volume 3 Issue 2; March-April-2016; Page No. 47-50

Fragmentation Techniques for Distribution Database: A Review

Sunil Kumar Verma
Himalayan University, Itanagar, Arunachal Pradesh (India)
iqrasoftware@gmail.com

ARTICLE INFO ABSTRACT

Received: 25 April. 2016 The wide acceptance of the relational approach in data-processing
Accepted 10 May 2016 applications and the continuous improvement of commercially existing
relational databases has increased the interest for using database in non-
Corresponding Author:
conventional applications, such as computer-aided design (CAD), geographic
Sunil Kumar Verma information systems, image, and graphic database systems. The Centralized
and distributed database systems are developed for balancing the load and
Himalayan University, Itanagar, scattering the data over different sites on an organization. So in order to
Arunachal Pradesh (India) distribute the database on different sites of an organization, fragmentation
Key words: Fragmentation, methods are used. There is several fragmentation methods reviewed in this
Centralized and Distributed paper.
Database System, ©2016, IJICSE, All Right Reserved
Hybrid fragmentation.

INTRODUCTION fragmentation strategy is chosen carefully according

to the database model. Fragments are made and then
Information storage has been a challenging endeavor
they are distributed to the sites where needed.
throughout human history and existed even before
Recent developments in network technologies have
modern computer systems [1]. The last three decades
increased the intensity of data in telecommunication
are marked by rapid growth of computer technology.
services. So the telecommunication sector is now at a
This has raised the needs to evolve new techniques to
point where they should move to fault-tolerant, more
manage huge amounts of data. Today, mostly
reliable and less expensive means of distributed data
centralized databases are used to store and manage
management [4]. But this emerging field has very
data [2]. They carry the advantages of high degree of
critical issues to be dealt with before implementation
security, concurrency and backup and recovery
in the telecom industry. Primarily these issues are:
control. However, they also have disadvantages of
concurrency control, availability, transparency,
high communication costs (when the client is far away
recovery and throughput etc. Compared with the use
and communication is very frequent), unavailability in
of distributed databases in other traditional
case of system failure and a single source bottleneck
industries, telecommunication databases need to
[3]. These issues raise the need of distribution of
fulfill some additional and tough requirements on
databases over various systems or locations. But the
availability, recovery, throughput, and reliability
main motivation behind the concept of distributed
generally and response time specifically [2]. In the
databases is the efficient management of huge
given context response time is of special importance
amount of data with increased availability and
before implementing distributed database at Ericsson.
reduced communication cost. Research conducted in
A lot of research has been done on distributed
1991 for distributed databases predicted a huge shift
databases in general; little is, however, done with
from traditional databases to distributed databases in
special consideration to the telecom environment [5].
the coming arena primarily due to organizational
In our thesis we present a state of the art report
needs to manage huge amounts of data [2]. According
about the response time and issues related to it
to Ilker, many applications in the future will be
including effect of increase in data and impact on the
distributed due to the development in technology,
response time when data is not locally present.
and therefore the databases will also be distributed
[3]. The telecommunication sector also wants to II. BACKGROUND THEORY
47

embrace this technology of data distribution. But

The purpose of this paper is to present some ground
before distribution, fragmentation is a very important
Page

knowledge to databases, their types and related

and critical task that needs to be done. The
 Sunil Kumar Verma / International Journal of Innovative Computer Science & Engineering

issues. A centralized database system is a system where all

the data is stored at a single location. Users retrieve
A. DATABASE SYSTEM:
the data from that database, perform some actions
The 20th century witnessed a lot of development in and store it again at that location. Making your
databases to meet the needs of information storage database centralized carry many advantages including
and retrieval. It started with simple file systems, easy data management, high degree of security,
stored individually at separate places. However, it was concurrency, and backup and recovery etc. The figure
not very fruitful for huge amounts of data due to 1 depicts the general architecture of centralized
problems of redundancy, separation and isolation of database system for two countries (A, B); all sites can
data, data dependence and incompatible file formats access the data from the central database server via
[6]. Later these files were merged into one unit and in their network. The obvious advantages of this system
1964 the term “database‟ was coined for this unit. include data consistency, security, and easy
Advantages associated with the database were faster management.
and shared access, data integrity and data consistency
C. DISTRIBUTED DATABASES:
etc. Later different database architectures were
devised according to the need of different
applications. IBM‟s IMS (Information Management
System) in 1966 was the first commercial hierarchical
database system [7]. It provided the features of easy
data updating, fast retrieval, multiple association and
simplicity of structure. However, pointer path restrict
access, data replication and large computer storage
made the scope of hierarchical databases very limited.
Object oriented database came out to support graph-
structured objects. They incorporated many useful
features of object oriented paradigms as
encapsulation, inheritance and object identity.
However their complexity of model and lack of
standard limited their use. One main problem with
these databases was the intermingling of conceptual Figure 2: Distributed Database system Architecture
relationships with the physical storage. Though it “A distributed database is a collection of multiple,
provided efficient access it reduced flexibility. The logically interrelated databases distributed over a
solution to this problem emerged in the inception of computer network” [2]. It may also be a single
relational databases by E.F.Codd in 1970 [7]. This database divided into chunks and distributed over
database model is based on the mathematical several locations. The database is scattered over
principles of set theory and predicate logic. It various locations which provides local access to data
provided faster access by joining multiple tables using and thus reduces communication costs and increases
relations. Relational databases are now present on availability. Most of today’s business applications
almost every type of computers. have shifted from traditional processing to online
B. CENTRALIZED DATABASES: processing.
Generally, distributed database is the collection of
databases distributed across different locations or
sites over a network as illustrated in Figure 2.
Similarly, it may also be a single database, divided into
chunks and distributed over various sites. Each site
has a certain amount of data that it needs frequently
and it can get the rest from some other site.
Distributed databases are very useful when
availability and fast response time is needed. They
increase performance and reduce communication
costs. However, many serious issues are also related
with distributed databases which must be
investigated. Among those issues, consistency,
48

concurrency control and query optimization.

Page

Figure 1: Centralized Database system Architecture

 Sunil Kumar Verma / International Journal of Innovative Computer Science & Engineering

III. FRAGMENTATION which are joined together are put on the same site. In
this thesis, derived horizontal fragmentation is used.
It is the decomposition of a relation into fragments
each being treated as a unit [2]. Fragmentation is C. HYBRID/MIXED FRAGMENTATION
done according to the data selection patterns of
applications running on the database. It permits to
divide a single query into a set of multiple sub-queries
that can execute parallel on fragments. Fragmentation
can be of any type: horizontal, vertical and
hybrid/mixed.
A. VERTICAL FRAGMENTATION:
It divides a relation into fragments which contain a
subset of attributes of a relation along with the
primary key attribute of the relation. The purpose of
vertical fragmentation is to partition a relation into a
set of smaller relations to enable user applications to
run on only one fragment [2].
Figure 5: Hybrid/Mixed Fragmentation

The Mixed/Hybrid fragmentation is Combination of

horizontal and vertical fragmentations. This type is
most complex one, because both types are used in
which horizontal and vertical fragmentation of the DB
application. The original relation is obtained back by
join or union operations.
The Mixed/Hybrid fragmentation is Combination of
Figure 3: Vertical Fragmentation horizontal and vertical fragmentations. This type is
B. HORIZONTAL FRAGMENTATION most complex one, because both types are used in
which horizontal and vertical fragmentation of the DB
It divides a relation into fragments along its tuples.
application. The original relation is obtained back by
Each fragment is a subset of tuples of a relation. It
join or union operations.
identifies some specific rows based on some criteria
and marks it as a fragment.

Figure 6: Mixed Fragmentation Chart

In this report, the data mining fragmentation

technique is used to improve the performance of the
distributed DW system and reduces the execution
time of queries. Here, allocation process and queries
Figure4: Horizontal Fragmentation process are also involved. The allocation process
Horizontal fragmentation is further divided into two allocates the data on the sites in network and
types. maintains the replication of data. Queries are used to
increase the accessing speed of data from the tables.
Primary horizontal fragmentation Fragmentation, Allocation and queries improve the
This type of fragmentation is done where the tables in efficiency and performance of the system.
a database are neither joined nor have dependencies. IV. CONCLUSION
So, no relationship exists among the tables.
It is important to manage an appropriate
Derived horizontal fragmentation methodology for data fragmentation in order to
49

Derived horizontal fragmentation is used for parent utilize the resources and thus it is must to select an
relation. It is used where tables are interlinked with accurate and efficient fragmentation methodology to
Page

the help of foreign keys. It ensures that the fragments enrich the power of distributed database system.
 Sunil Kumar Verma / International Journal of Innovative Computer Science & Engineering

REFERENCES VLDB-99, Edinburgh, Scotland, UK, September 6th

1999 : proceedings. Springer, Berlin.
1. Ebrary Inc. 1999. Funding a revolution [Elektronisk
5. Cooper, B.F., Ramakrichnan, R., Srivastava, U.,
resurs] government support for computing
Silberstein, A., Bohannon, P., Jacobsen, H.-A., Puz,
research. National Academy Press, Washington,
N.,Weaver, D. And Yerneni, R. August 2008.
D.C.
PNUTS: Yahoo!'s hosted data serving platform.
2. Özsu, M.T. And Valduriez, P. 1999. Principles of
6. Connolly, T.M. And Begg, C.E. 1998. Database
distributed database systems. Prentice Hall, Upper
systems: a practical approach to design,
Saddle River, N.J.
implementation and management. Addison-
3. Köse , I. Spring , 2002. Distributed Database
Wesley, Harlow.
Security.
7. Vaughn 14 Nov, 2003. CSPC 343: A Sketch of
4. Jonker, W. 2000. Databases in telecommuni-
Database History.
cations: international workshop co- located with

50
Page