BGP AS question

[mahesh parmar]

Hi all,

i was reading about BGP and it says ----

It is important to remember that when a BGP speaker receives an update from other BGP speakers in its own AS (iBGP), the receiving BGP speaker will not redistribute that information to other BGP speakers in its own AS. The receiving BGP speaker will redistribute that information to other BGP speakers outside of its AS. That is why it is important to sustain a full mesh between the iBGP speakers within an AS.

I need to know why receiving BGP speaker will not redistribute that info to other BGP speakers in its own AS? if someone can explain me this in detail.

Also when it says that it is important to sustain a full mesh between iBGP speakes within an AS  does ot mean that all Routers in AS  must have

physical connection to all other routers in that AS  to provide redundancy.?

Thanks

MAhesh

[Reza Sharifi · Accepted solution]

Mahesh,

Fully mesh IBGP is required, because when an IBGP speaker seen its own AS number in the update it will ignore the update, if not it will cause routing loop.  There are ways around this by using a route reflectors or confederations.

Also when it says that it is important to sustain a full mesh between iBGP speakes within an AS  does ot mean that all Routers in AS  must have

physical connection to all other routers in that AS  to provide redundancy.?

no, only logical connectivity is required.  So if you have 3 routers physically connected as below

r1---------r2---------r3

you need to have an IBGP peer from r1 to r2 and to r3 and from r2 to r1 and to r3 and from r3 to r2 and to r1.

For BGP RR

http://www.rfc-editor.org/rfc/rfc4456.txt

HTH

Reza

[abhisar patil · Accepted solution]

While desiging BGP it is considered that there is Full Mesh connectivity inside one AS.So any router will not

send routes received from one router to another as the other router must have connectivity with original router(as full mesh) and it has routes of original router.It is to avoid loops.

But practically it is not possible to have full mesh across a big network.So solution is Route Reflector and Confederations.

And for iBGP peers direct connection is not required between two routers but it is required in eBGP peering.We can form logical link between routers in iBGP like virtual link in OSPF.

Abhisar.

[Giuseppe Larosa · Accepted solution]

Hello Mahesh,

a BGP speaker is not allowed to propagate a BGP advertisement received from an iBGP peer to other iBGP peers.

This is done to avoid routing loops within an AS topology.

The reason is that when the BGP advertisement is received by an iBGP peer the AS path attribute is left unchanged so it is not possible to know if the BGP advertisement has travelled on one or more iBGP peers it is not traced.

To overcome this protocol safety measure you need an iBGP full mesh that is on every BGP speaker you should define an iBGP session to any other iBGP speaker in the same AS

The requirement is not to have a full mesh of physical links but of BGP sessions.

However, even if the requirement is only for logical sessions the iBGP full mesh is a scalability challange.

Scalability in iBGP is achieved by using:

BGP route reflector servers

and /or BGP confederations.

These two tools work by adding BGP attributes that allow for safe propagation of BGP advertisements within an AS:

BGP route reflector servers add a Cluster -list attribute + Originator-ID.

Originator-Id = BGP router-id of BGP speaker that originated the advertisement

Cluster-list = list of RRS that have reflected the advertisement can be made of BGP router-ids or cluster-ids

BGP confederations create a parallel AS path confederation attribute that traces the mini AS that have seen the advertisement.

A BGP router knows what are the other mini AS and so it discriminates between true eBGP sessions to outer word and confederation eBGP sessions to routers in other mini AS

To outside worid the official AS number is used and mini AS are not visible in other AS.

A BGP RRS is allowed to reflect what has received from a BGP client to other clients and non clients.

the two tools can be combined in very large service provider networks with RRS used within each mini AS in BGP confederations.

Hope to help

Giuseppe