BGP

BGP
4-1
Copyright 2005 Juniper Networks, Inc.
Proprietary and Confidential
www.juniper.net
BGP Operation
BGP stores routes in three main routing information
base (RIB) memory tables
Adjacency-RIB-IN: Contains all received routes from each

peer
RIB-LOCAL: Contains routes the local router uses to forward
traffic
Adjacency-RIB-OUT: Contains all advertised routes sent to
each peer
Only active BGP routes in the local routing table can

be advertised to peers
Single, best BGP path is advertised
Can use advertise-inactive when BGP route is not
active, but only the single, best, inactive BGP path is
advertised
2008 Juniper Networks, Inc. All rights reserved.
BGP Update Message

Advertised and withdrawn routes are sent in an
update message to established peers
Common BGP header fields:
(16-byte) Marker
(2-byte) Length
(1-byte) Type
Update message fields:

(2-byte) Unfeasible routes length
(Variable) Withdrawn routes
(2-byte) Total path attributes length
(Variable) Path attributes
(Variable) Network layer reachability information
Local Preference
Often used to set the exit point from an autonomous
system
Only exchanged within the confines of a single AS
(1-byte) Attribute type
(Variable) Attribute length
(4-byte) Local preference value
AS Path
Collection of autonomous system numbers through
which the route has passed
Used to prevent routing loops
Used to select an active BGP path
(Variable) Attribute length
(Variable) AS path segments
Origin
Measure of believability set by the originating BGP
router
Internal (0) > External (1) > Incomplete (2)
JUNOS software sets all injected routes to Internal
(1-byte) Attribute length
(1-byte) Origin
Multiple Exit Discriminator

Often used to set an incoming entry point for an AS
Never passed through one AS to another AS
Can be translated from IGP metric
(1-byte) Attribute length
(4-byte) Multiple exit discriminator
BGP Path Selection Process (1 of 2)

Steps:
1. Can the BGP next hop be reached?
If yes, proceed
If no, stop processing
2. Prefer the highest local preference value
3. Prefer the shortest AS path length
4. Prefer the lowest origin value
5. Prefer the lowest MED value

6. Prefer paths learned via EBGP over routes learned via IBGP
BGP Path Selection Process (2 of 2)

7. Prefer paths with the lowest IGP metric. Select a physical next
hop (or hops) for each peer
a. Examine routing tables inet.0 and inet.3 for the BGP next hop.
Install the physical next hop(s) for the route with the better preference
b. In the event of a preference tie, install the physical next hop(s) found
in inet.3
c. For preference ties within the same routing table, install the physical
next hop(s) where the greater number of equal-cost paths exist
8. Prefer paths with the shortest cluster list length

9. Prefer paths from the peer with the lowest RID
a. When routes from different AS are equal up to the RID comparison
stage, prefer currently active route
10.Prefer paths from the peer with the lowest peer ID
Local Preference Selection

user@host> show route 192.168.48/24 detail
inet.0: 52 destinations, 94 routes (52 active, 0 holddown, 0 hidden)
192.168.48.0/24 (3 entries, 1 announced)
*BGP
Preference: 170/-201
Source: 192.168.32.1
Next hop: 10.222.28.2 via fe-0/0/0.0, selected
Protocol next hop: 192.168.32.1 Indirect next hop: 858b4e0 73
State: <Active Int Ext>
Local AS: 65432 Peer AS: 65432
Age: 18:57
Metric2: 3
Task: BGP_65432.192.168.32.1+1042
AS path: 65000 I
Localpref: 200
Router ID: 192.168.32.1
BGP
Source: 10.222.29.2
Next hop: 10.222.29.2 via ge-0/1/0.0, selected
State: <Ext>
Inactive reason: Local Preference
Age: 19:09
Task: BGP_65000.10.222.29.2+1061
AS path: 65000 I
---(more)---
10
AS Path Selection
*BGP
Source: 192.168.16.1
Protocol next hop: 192.168.16.1 Indirect next hop: 85d8270 46
Age: 3:41
Metric2: 3
Task: BGP_65432.192.168.16.1+179
AS path: 65000 I
Localpref: 100
Router ID: 192.168.16.1
BGP
Source: 10.222.45.1
Next hop: 10.222.45.1 via so-0/3/0.0, selected
State: <Ext>
Inactive reason: AS path
Age: 25
Task: BGP_65000.10.222.45.1+1064
AS path: 65000 65000 65000 I
---(more)---
11
Origin Selection
*BGP
Source: 10.222.6.1
State: <Active Ext>
Age: 1:29
Metric: 20
Task: BGP_65000.10.222.6.1+1084
AS path: 65000 I
Localpref: 100
Router ID: 192.168.48.1
BGP
Source: 10.222.45.1
State: <Ext>
Inactive reason: Origin
Age: 53
Task: BGP_65000.10.222.45.1+1064
AS path: 65000 ?
Localpref: 100
---(more)---
12
MED Selection
*BGP
Source: 192.168.16.1
Age: 9:53
Metric: 5
Metric2: 3
Task: BGP_65432.192.168.16.1+179
AS path: 65000 I
Localpref: 100
Router ID: 192.168.16.1
BGP
Source: 10.222.45.1
State: <NotBest Ext>
Inactive reason: Not Best in its group
Age: 2:21
Metric: 20
Task: BGP_65000.10.222.45.1+1064
AS path: 65000 I
---(more)---
13
EBGP over IBGP Selection

*BGP
Source: 10.222.5.2
State: <Active Ext>
Age: 2:18
Task: BGP_65000.10.222.5.2+1067
AS path: 65000 I
Localpref: 100
Router ID: 192.168.52.1
BGP
Source: 192.168.32.1
Protocol next hop: 192.168.32.1 Indirect next hop: 858b4e0 73
State: <NotBest Int Ext>
Inactive reason: Interior > Exterior > Exterior via Interior
Age: 5:25
Metric2: 3
Task: BGP_65432.192.168.32.1+1042
AS path: 65000 I
---(more)---
14
IGP Cost Selection

*BGP
Source: 192.168.16.1
Protocol next hop: 192.168.16.1 Indirect next hop: 85d61d4 41
Age: 12
Metric2: 1
Task: BGP_65432.192.168.16.1+1126
AS path: 65000 I
Localpref: 100
Router ID: 192.168.16.1
BGP
Source: 192.168.32.1
Protocol next hop: 192.168.32.1 Indirect next hop: 85d630c 44
Inactive reason: IGP metric
Age: 3:19
Metric2: 2
Task: BGP_65432.192.168.32.1+1045
---(more)---
15
Cluster List Length Selection

*BGP
Age: 6:41
Metric: 20
Metric2: 3
AS path: 65432 I (Originator) Cluster list: 3.3.3.3
AS path: Originator ID: 192.168.24.1
Localpref: 100
Router ID: 192.168.36.1
BGP
Protocol next hop: 192.168.20.1 Indirect next hop: 85d83a8 72
Inactive reason: Cluster list length
Age: 1:13
Metric: 20
Metric2: 3
AS path: 65432 I (Originator) Cluster list: 2.2.2.2 1.1.1.1
AS path: Originator ID: 192.168.20.1
Localpref: 100
Router ID: 192.168.48.1
16
Router ID Selection
*BGP
Source: 10.222.29.2
State: <Active Ext>
Age: 10
Task: BGP_65000.10.222.29.2+179
AS path: 65000 I
Localpref: 100
Router ID: 192.168.20.1
BGP
Source: 10.222.5.2
Inactive reason: Router ID
Task: BGP_65000.10.222.5.2+1067
AS path: 65000 I
Localpref: 100
Router ID: 192.168.52.1
---(more 66%)---
17
Peer ID Selection
*BGP
Source: 10.222.5.2
State: <Active Ext>
Age: 30
Task: BGP_65000.10.222.5.2+1073
AS path: 65000 I
Localpref: 100
Router ID: 192.168.52.1
BGP
Source: 10.222.15.2
Inactive reason: Update source
Task: BGP_65000.10.222.15.2+179
AS path: 65000 I
Localpref: 100
Router ID: 192.168.52.1
---(more 51%)---
18
Path Selection: Peer ID

Used when multiple BGP peering sessions exist
between two routers
Only one BGP session is used to forward traffic
ID is the physical IP address on the neighboring routers
interface
Los Angeles
10.10.1.2/24
10.10.1.1/24
(AS1)
10.10.2.2/24
10.10.2.1/24
Las Vegas
(AS2)
[edit protocols bgp group ext-peers]

type external;
peer-as 2;
neighbor 10.10.1.1;
neighbor 10.10.2.1;
}
19
BGP Multipath (1 of 3)
BGP can ignore both router ID and peer ID
comparisons when multipath is configured within
BGP
Can use:
Two peering sessions to the same router

Two peering sessions to different routers in the same AS
Los Angeles
(AS1)
10.222.28.1/24
10.222.29.1/24

type external;
peer-as 2;
neighbor 10.222.28.2;
neighbor 10.222.29.2;
}
user@host> show bgp summary
Peer
AS
InPkt
OutPkt
10.222.28.2
2
7
7
10.222.29.2
2
8
10
10.222.28.2/24
Las Vegas
(AS2)
Reno
10.222.29.2/24
OutQ
0
0
(AS2)
Flaps Last Up/Dwn State|#Active/Rec

0
00:00:02 4/4/0
0
00:00:06 0/4/0
20
Routes from each peer contain a single next hop
user@host> show route protocol bgp terse
+ = Active Route, - = Last Active, * = Both
A Destination
P Prf
Metric 1
* 172.16.20.4/30
B 170
100
>10.222.28.2
2 I
B 170
100
>10.222.29.2
2 I
B 170
100
>10.222.28.2
2 I
B 170
100
>10.222.29.2
2 I
B 170
100
>10.222.28.2
2 I
B 170
100
>10.222.29.2
2 I
B 170
100
>10.222.28.2
2 I
B 170
100
>10.222.29.2
2 I
* 172.16.20.8/30
* 172.16.20.12/30
* 172.16.20.16/30
Metric 2
Next hop
AS path
21
Peer group on Los Angeles configured with
multipath
Active route receives two next hops
Forwarding table still maintains a single next hop per route
user@host> show route protocol bgp terse
A Destination
* 172.16.20.4/30
P Prf
B 170
Metric 1
100
* 172.16.20.8/30
B 170
B 170
100
100
* 172.16.20.12/30
B 170
B 170
100
100
* 172.16.20.16/30
B 170
B 170
100
100
B 170
100
Metric 2
Next hop
>10.222.28.2
10.222.29.2
>10.222.29.2
>10.222.28.2
10.222.29.2
>10.222.29.2
>10.222.28.2
10.222.29.2
>10.222.29.2
10.222.28.2
>10.222.29.2
>10.222.29.2
AS path
2 I
2 I
2 I
2 I
2 I
2 I
2 I
2 I
22
Multihop Peering
EBGP sessions can peer with nonphysical addresses
lo0: 192.168.3.4
lo0: 172.16.128.1
Los Angeles
10.10.1.2/24
10.10.1.1/24
(AS1)
10.10.2.2/24
10.10.2.1/24
Las Vegas
(AS2)

type external;
Step 1
local-address 192.168.3.4;
neighbor 172.16.128.1 {
A TTL value of 1 accommodates peering to
Step
2
multihop ttl 1;
a loopback address on a directly
connected peerhigher values are needed
}
for peers that are not directly connected
[edit routing-options]
static {
Step 3
route 172.16.128.1 next-hop [ 10.10.1.1 10.10.2.1 ];
}
23
Multiple Next Hops

Both multihop and multipath create routes with
multiple next hops in the routing table
Use a routing policy to forward traffic on both next hops
user@host> show route 172.16.20.4/30 terse
A Destination
* 172.16.20.4/30
P Prf
B 170
Metric 1
100
B 170
100
Metric 2
user@host> show route forwarding-table matching

Routing table: inet
Internet:
Destination
Type RtRef Next hop
172.16.20.4/30
user
0
10.10.1.1
Next hop
>10.10.1.1
10.10.2.1
>10.10.2.1
AS path
2 I
2 I
172.16.20.4/30
Type Index NhRef Netif

indr
106
4
ucst
47
4 fe-0/0/0.0
24
Load Balancing
user@host> show configuration policy-options policy-statement load-balance
then {
load-balance per-packet;
}
user@host> show configuration routing-options forwarding-table
export load-balance;
user@host> show route 172.16.20.4/30 terse
A Destination
* 172.16.20.4/30
P Prf
B 170
Metric 1
100
B 170
100
Metric 2
Next hop
>10.10.1.1
10.10.2.1
>10.10.2.1
AS path
2 I
2 I
user@host> show route forwarding-table matching 172.16.20.4/30

Routing table: inet
Internet:
Destination
Type RtRef Next hop
Type Index NhRef Netif
172.16.20.4/30
user
0
ulst
108
4
indr
106
2
10.10.1.1
ucst
47
4 fe-0/0/0.0
indr
107
2
10.10.2.1
ucst
99
4 ge-0/1/0.0
25
Multihop with No Next-Hop Change (1 of 4)

AS 65001 and 65002 wants to use MPLS for
forwarding user traffic
LSP from Hong Kong to Amsterdam is one example
The 172.16.1.0/24 route is advertised into AS 65002

by an EBGP peer
172.16.1.0/24
via EBGP
AS 65001
AS 65002
LSP
Hong Kong
192.168.16.1
San Jose
192.168.20.1
Montreal
192.168.40.1
Amsterdam
192.168.32.1
26

The 172.16.1.0/24 route currently does not use the
LSP to Amsterdam
BGP next hop is not in the inet.3 routing table
lab@HongKong> show route table inet.3
192.168.32.1/32
*[RSVP/7] 00:14:02, metric 0

> to 10.222.29.2 via ge-0/1/0.0, label-switched-path HK-to-Amsterdam
lab@HongKong> show route 172.16/16 detail

172.16.1.0/24 (1 entry, 1 announced)
*BGP
Source: 192.168.20.1
Protocol next hop: 192.168.40.1 Indirect next hop: 858709c 119
Age: 26:02
Metric2: 0
Task: BGP_65001.192.168.20.1+3187
Announcement bits (2): 0-KRT 2-Resolve inet.0
AS path: 65002 65003 I
Localpref: 100
Router ID: 192.168.20.1
27

EBGP peers alter the BGP next hop across AS
boundaries
The show route advertising-protocol bgp
command has a next-hop value of self
If you use multihop and the keyword, no-nexthopchange, you can change default
The next-hop value is the address of the IBGP peer that first
advertised the route
user@Montreal> show route advertising-protocol bgp 192.168.20.1
Prefix
Nexthop
MED
Lclpref
AS path
* 172.16.1.0/24
Self
65003 I
user@Montreal# set neighbor 192.168.20.1 multihop no-nexthop-change
user@Montreal> show route advertising-protocol bgp 192.168.20.1
Prefix
Nexthop
MED
Lclpref
AS path
* 172.16.1.0/24
192.168.32.1
65003 I
28

User traffic uses the LSP to reach 172.16.1.0/24
user@HongKong> show route table inet.3
192.168.32.1/32
*[RSVP/7] 01:21:26, metric 0

user@HongKong> show route 172.16/16

172.16.1.0/24
*[BGP/170] 01:33:20, localpref 100, from 192.168.20.1

AS path: 65002 65003 I
user@HongKong> traceroute 172.16.1.1 source 192.168.16.1

traceroute to 172.16.1.1 (172.16.1.1) from 192.168.16.1, 30 hops max, 40 byte packets
1 10.222.29.2 (10.222.29.2) 1.114 ms 0.913 ms 0.831 ms
MPLS Label=100010 CoS=0 TTL=1 S=1
2 10.222.1.2 (10.222.1.2) 0.916 ms 0.868 ms 0.851 ms
MPLS Label=100008 CoS=0 TTL=1 S=1
3 10.222.45.2 (10.222.45.2) 0.682 ms 0.687 ms 0.641 ms
4 10.222.6.1 (10.222.6.1) 0.734 ms 0.739 ms 0.690 ms
29
Peer Configuration Options (1 of 3)

passive keeps BGP from sending open message
[edit protocols bgp]
group ext-peers {
type external;
peer-as 2;
neighbor 10.10.10.1 {
passive;
}
}
allow accepts open messages from any peer within

the configured IP address range
group ext-peers {
type external;
peer-as 65000;
allow 10.10/16;
}
Can enable MD5 authentication

group ext-peers {
type external;
peer-as 2;
neighbor 10.10.10.1 {
authentication-key $9$.mQn/9pBRSAp7VYojiAp0O1h;
}
}
30

prefix-limit allows a specified amount of
prefixes to be received
group ext-peers {
type external;
peer-as 2;
family inet {
unicast {
prefix-limit {
maximum 25000;
teardown 80 idle-timeout 10;
}
}
}
neighbor 10.10.10.1;
}
hold-time alters the value used in the session

negotiation process
group ext-peers {
type external;
hold-time 45;
peer-as 2;
}
31

advertise-peer-as disables suppression of route
advertisements
Routes learned from an EBGP peer are advertised back to
the same EBGP peer
Routes learned from an EBGP peer are advertised to EBGP
peers in the same AS as the originating peer
group ext-peers {
advertise-peer-as;
peer-as 2;
}
32
Graceful Restart
A BGP router can negotiate with peers to maintain its
announced routes during a restart
Neighbors continue forwarding traffic to the router
Neighbors do not withdraw the routes of the restarting router
End-of-RIB markers sent for each NLRI

Notifies the neighbor that all current routing information was
sent
Local router defers path selection algorithm until the marker
is received
Configured globally within [edit routingoptions]
33
Modifying Local Preference

AS 65002
AS 65003
192.168.19.0/24
local-preference
user@host> show route advertising-protocol bgp 192.168.40.1
Restart Complete
Prefix
Nexthop
MED
Lclpref
AS path
* 192.168.19.0/24
Self
0
100
65003 I
[edit]
user@host# set protocols bgp group int-peers local-preference 300
user@host> show route advertising-protocol bgp 192.168.40.1

Restart Complete
Prefix
Nexthop
MED
Lclpref
AS path
* 192.168.19.0/24
Self
0
300
65003 I
34
Modifying AS Path: remove-private

192.168.17.0/24: 1000
192.168.18.0/24: 1000
192.168.19.0/24: 1000
remove-private
AS 1000
192.168.17.0/24: 65001
192.168.18.0/24: 65002
192.168.19.0/24: 65003
AS 65001
AS 65002
AS 65003
192.168.17.0/24
192.168.18.0/24
192.168.19.0/24
35
Modifying AS Path: local-as (1 of 3)
172.16.10.0/24: 1 222
Internet
172.16.12.0/24: 1 333
AS 1
172.16.10.0/24: 222
AS 222
172.16.10.0/24
172.16.12.0/24: 333
AS 333
172.16.12.0/24
36

AS 777
172.16.10.0/24: 1 222
172.16.12.0/24: 1 333
172.16.10.0/24: 777 1 222
172.16.12.0/24: 777 1 333

local-as 1
172.16.10.0/24: 222
Internet
172.16.12.0/24: 333
AS 222
AS 333
172.16.10.0/24
172.16.12.0/24
37

AS 777
172.16.10.0/24: 222
172.16.12.0/24: 333
172.16.10.0/24: 777 222

172.16.12.0/24: 777 333
local-as 1 private
172.16.10.0/24: 222
Internet
172.16.12.0/24: 333
AS 222
AS 333
172.16.10.0/24
172.16.12.0/24
38
Modifying AS Path: as-override

172.16.10.0/24: 65022
172.16.10.0/24: 65432 65432

10.222.4.2
AS 65022
AS 65022
10.222.4.1
172.16.10.0/24
AS 65432
as-override
user@AS65432> show route advertising-protocol bgp 10.222.4.2
Prefix
Nexthop
MED
Lclpref
* 172.16.10.0/24
Self
AS path
65022 I
user@AS65022> show route receive-protocol bgp 10.222.4.1

[edit]
user@AS65432# set protocols bgp group AS-65022 as-override
user@AS65432> show route advertising-protocol bgp 10.222.4.2
Prefix
Nexthop
MED
Lclpref
* 172.16.10.0/24
Self
AS path
65022 I

Prefix
Nexthop
MED
Lclpref
* 172.16.10.0/24
10.222.4.1
AS path
65432 65432 I
39
Modifying AS Path: loops

172.16.10.0/24: 65022
AS 65022
172.16.10.0/24
172.16.10.0/24: 65432 65022
AS 65432
AS 65022
[edit]
user@AS65022# set routing-options autonomous-system 65022 loops 2

Prefix
Nexthop
MED
Lclpref
* 172.16.10.0/24
10.222.4.1
AS path
65432 65022 I
40
Coordinating MED and IGP Metrics

Use the metric-out command with a group or
neighbor
Can be set to:
A specific value
The current IGP metric
The minimum IGP metric ever learned
Can add to or subtract from the IGP metric

group as-100-peers {
type external;
peer-as 100;
neighbor 192.168.2.2
}
metric-out
metric-out
metric-out
metric-out
10;
igp;
minimum-igp;
igp 5;
41
Path Selection and MEDs

By default, the JUNOS software uses a deterministic
MED comparison scheme for routes from the same AS
always-compare-med compares MED values,
regardless of whether the neighboring AS is the same
Use with cautionevery network has a different
interpretation of a good MED
cisco-non-deterministic compares paths

based on when they are received
Not recommended for use in your network
Can cause incorrect route selections
[edit]
user@host# set protocols bgp path-selection ?
Possible completions:
always-compare-med
Always compare MED values, regardless of neighbor AS
cisco-non-deterministic Use Cisco IOS nondeterministic path selection algorithm
42

BGP

Uploaded by

Copyright:

Available Formats

BGP

Uploaded by

Document Information

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

BGP

Uploaded by

Copyright:

Available Formats

BGP

Proprietary and Confidential

Adjacency-RIB-IN: Contains all received routes from each

Only active BGP routes in the local routing table can

2008 Juniper Networks, Inc. All rights reserved.

BGP Update Message

Update message fields:

2008 Juniper Networks, Inc. All rights reserved.

2008 Juniper Networks, Inc. All rights reserved.

2008 Juniper Networks, Inc. All rights reserved.

2008 Juniper Networks, Inc. All rights reserved.

Multiple Exit Discriminator

2008 Juniper Networks, Inc. All rights reserved.

BGP Path Selection Process (1 of 2)

5. Prefer the lowest MED value

2008 Juniper Networks, Inc. All rights reserved.

BGP Path Selection Process (2 of 2)

8. Prefer paths with the shortest cluster list length

10.Prefer paths from the peer with the lowest peer ID

2008 Juniper Networks, Inc. All rights reserved.

Local Preference Selection

2008 Juniper Networks, Inc. All rights reserved.

2008 Juniper Networks, Inc. All rights reserved.

2008 Juniper Networks, Inc. All rights reserved.

2008 Juniper Networks, Inc. All rights reserved.

EBGP over IBGP Selection

2008 Juniper Networks, Inc. All rights reserved.

IGP Cost Selection

2008 Juniper Networks, Inc. All rights reserved.

Cluster List Length Selection

2008 Juniper Networks, Inc. All rights reserved.

2008 Juniper Networks, Inc. All rights reserved.

2008 Juniper Networks, Inc. All rights reserved.

Path Selection: Peer ID

[edit protocols bgp group ext-peers]

Two peering sessions to the same router

[edit protocols bgp group ext-peers]

Flaps Last Up/Dwn State|#Active/Rec

2008 Juniper Networks, Inc. All rights reserved.

2008 Juniper Networks, Inc. All rights reserved.

[edit protocols bgp group ext-peers]

2008 Juniper Networks, Inc. All rights reserved.

Multiple Next Hops

user@host> show route forwarding-table matching

2008 Juniper Networks, Inc. All rights reserved.

Type Index NhRef Netif

user@host> show route forwarding-table matching 172.16.20.4/30

2008 Juniper Networks, Inc. All rights reserved.

Multihop with No Next-Hop Change (1 of 4)

The 172.16.1.0/24 route is advertised into AS 65002

2008 Juniper Networks, Inc. All rights reserved.

Multihop with No Next-Hop Change (2 of 4)

*[RSVP/7] 00:14:02, metric 0

lab@HongKong> show route 172.16/16 detail

2008 Juniper Networks, Inc. All rights reserved.

Multihop with No Next-Hop Change (3 of 4)

Multihop with No Next-Hop Change (4 of 4)

*[RSVP/7] 01:21:26, metric 0

user@HongKong> show route 172.16/16

*[BGP/170] 01:33:20, localpref 100, from 192.168.20.1