Scribd
Upload
74 views13 pages

iBGP Design Solutions and Route Reflectors

This document discusses different solutions for iBGP design in large networks: 1. Full mesh iBGP peering allows all routers to learn paths but does not scale well. 2. Route reflectors eliminate the full mesh requirement by having clients peer only to route reflectors, which then advertise routes. Route reflectors use mechanisms like the CLUSTER_LIST to prevent routing loops. 3. Confederations split an autonomous system into sub-ASNs to reduce the number of iBGP sessions needed at the cost of complexity. External peers treat the confederation as a single AS.

Uploaded by

Meh Di
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
74 views13 pages

iBGP Design Solutions and Route Reflectors

This document discusses different solutions for iBGP design in large networks: 1. Full mesh iBGP peering allows all routers to learn paths but does not scale well. 2. Route reflectors eliminate the full mesh requirement by having clients peer only to route reflectors, which then advertise routes. Route reflectors use mechanisms like the CLUSTER_LIST to prevent routing loops. 3. Confederations split an autonomous system into sub-ASNs to reduce the number of iBGP sessions needed at the cost of complexity. External peers treat the confederation as a single AS.

Uploaded by

Meh Di
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

CCIE DATA CENTER iBGP Design Solutions

BGP

CREATED BY SALMAN ALHIARY, CCIE #56363


iBGP Design Solutions Overview
• iBGP performs loop prevention via route filtering. iBGP learned routes cannot
be advertised to another iBGP neighbor.

• This implies iBGP neighborship requires one of the following solutions:


• Fully meshed iBGP peering.
• Confederations.
• Route reflectors (RR).

CREATED BY SALMAN ALHIARY, CCIE #56363


iBGP Full Mesh

• Full mesh design advantages:


• Path diversity; all BGP peers learn all possible egress paths.
• Optimal traffic flows; all BGP peers learn the closest egress path.
• Full mesh design disadvantages:
• Control plane scaling is exponential. Full mesh means n*(n-1)/2 peerings, where n is the
number of routers in the autonomous system.
• 10 routers require 45 peerings.
• 100 routers require 4950 peerings.

CREATED BY SALMAN ALHIARY, CCIE #56363


BGP Sub-ASN 65001
1 Update: 21.0.0.0/8
AS_PATH: 45

Confederation 2

• BGP confederation, separates


each router in the AS into one of 21.0.0.0/8 3
3 21.0.0.0/8
several confederation sub-AS. AS_PATH: (65001) 45 AS_PATH: (65001) 45

• Peers inside the same sub-AS are


Sub-ASN 65002 Sub-ASN 65003
confederation iBGP peers.
• Routers in different sub-AS are 4
4
confederation eBGP peers.
• Inside sub-ASs, full mesh or RR 6
requirement remains.
5 5 21.0.0.0/8
• Between sub-ASs, routers act like 21.0.0.0/8 AS_PATH: 123 45
normal eBGP neighbors; i.e., they AS_PATH: (65002 65001) 45
can advertise iBGP routes learned
inside their confederation sub-AS
into another confederation sub- • Devices outside the confederation do not know about the internal structure!! Sub-AS numbers are
stripped from advertisements to “true” eBGP peers.
AS.
• BGP routers in a confederation add the subautonomous systems (sub-AS) into the AS_PATH as part
• Typically, confederations use ASNs of an AS_PATH segment called the AS_CONFED_SEQ.
in private range (64512 – 65534).
CREATED BY SALMAN ALHIARY, CCIE #56363
iBGP Route Reflection (RR)
• RR eliminates the need for full mesh design. Only peering to the RR needed.
• Like OSPF DR and IS-IS DIS, RR minimizes prefix replication by:
• Send one update to the RR.
• RR sends the update to its “clients”.
• RR does not modify prefix’s attributes when reflecting routes. (the next hop attribute
will not be changed even with the next-hop-self command).
• RR is per Address Family. IPv4 RR doesn’t not imply IPv6 RR.
• Any client of the first RR could be a RR for other routers downstream.
• Route Reflector Peerings:
• eBGP peers.
• iBGP Client Peers.
• iBGP Non-Client Peers.

CREATED BY SALMAN ALHIARY, CCIE #56363


RR Update & Non-Clients

Loop Prevention 4
RR Cluster 5
• Only the RR uses the following 4
rules! the other routers (clients RR
Server
and non-clients) are not even
2 Non-Clients
aware of the RR: 3
• eBGP learned routes pass to eBGP
peers, iBGP Clients peers, & iBGP
Non-Clients peers.
RR Clients
21.0.0.0/8 1

CREATED BY SALMAN ALHIARY, CCIE #56363


RR Update & Non-Clients

Loop Prevention 4
RR Cluster 5
• Only the RR uses the following 4
rules! the other routers (clients RR
Server
and non-clients) are not even
11.0.0.0/8 Non-Clients
aware of the RR: 2
• eBGP learned routes pass to eBGP 1
peers, iBGP Clients peers, & iBGP
Non-Clients peers.
• iBGP Client learned routes pass to 3 RR Clients
eBGP peers, iBGP Clients peers, &
iBGP Non-Clients peers.

CREATED BY SALMAN ALHIARY, CCIE #56363


RR Update & 12.0.0.0/8
Non-Clients

1
Loop Prevention
RR Cluster
• Only the RR uses the following 4
rules! the other routers (clients RR
Server
and non-clients) are not even
Non-Clients
aware of the RR: 2
• eBGP learned routes pass to eBGP
peers, iBGP Clients peers, & iBGP
Non-Clients peers.
• iBGP Client learned routes pass to 3 RR Clients
eBGP peers, iBGP Clients peers, &
iBGP Non-Clients peers.
• iBGP Non-Client learned routes
pass to eBGP peers, iBGP Clients
peers, but not to iBGP Non-Client
peers.

CREATED BY SALMAN ALHIARY, CCIE #56363


iBGP Route Reflection (Cont.)
• The RR feature uses several tools to prevent loops, as follows:
• CLUSTER_LIST: RRs add their cluster-ID into a BGP PA called the CLUSTER_LIST before
sending prefix Updates.
• When receiving a BGP Update, RRs discard received prefixes for which their cluster-ID
already appears in the cluster-list. This prevents RRs from looping advertisements
between clusters.
• The CLUSTER_LIST is a sequential list just like AS_SEQ.
• ORIGINATOR_ID: This PA lists the BGP-RID of the first iBGP peer who advertised the
route into the AS (added by the RR).
• If a router sees its own BGP-RID as the ORIGINATOR_ID in a received route, it will not
use or propagate the route.
• RR advertises the best routes ONLY. RRs reflect routes only if the RR considers the
route to be a “best” route in its own BGP table.

CREATED BY SALMAN ALHIARY, CCIE #56363


Large Scale Route
Reflection

CREATED BY SALMAN ALHIARY, CCIE #56363


Route Reflector Configuration
• The RR feature configured on the RR server only! No configuration in RR
Clients:
• Step 1: if needed change the default cluster-id (equals to RID).
• Step 2: Set the RR client in RR server.

router bgp 65000


cluster-id 1.2.3.4
neighbor 10.1.1.10
address-family ipv4 unicast
route-reflector-client

CREATED BY SALMAN ALHIARY, CCIE #56363


RR Server

RR Client RR Client

CREATED BY SALMAN ALHIARY, CCIE #56363


Thanks for watching!

CREATED BY SALMAN ALHIARY, CCIE #56363

You might also like