Hello All,
I was reading the CCIE Routing and Switching Cert Guide V5 and came across the statement: iBGP peers do not share routes with each other in case of a routing loop, thus, all iBGP peers must form a full mesh within an AS. (without the confederation or router reflector)
I understand that iBGP doesn't by default come with a routing loop prevention mechanism like AS_PATH checks. But this confuses me a little about what's the function of iBGP router. Is there any other usage of an iBGP router than the following:
- Work as a border router between IGP and BGP in the same AS, introduce redistributed routes from IGP
- Work with other iBGP routers to share roughly the same eBGP Update message across the AS
And finally, for the 2nd point, I don't know exactly how it works so that all iBGP routers in the same AS gets the same eBGP update. what's the difference of handling the Update message other than it doesn't add the local AS number to AS_PATH?
---------------------------------------2nd edit here -----------------------------------
Please allow me to clarify my question a little:
Yes I understand why we need iBGP. Because we want multiple routers in the same AS and share the AS number.
What I don't understand, is the mechanism of it sharing the routes to the BGP instance, in particular, towards other iBGP peers from IGP redistribution.
If the route came from eBGP, I'd assume it will certainly forward the BGP Update message towards its iBGP peer, just don't add another AS number and it will get accepted and updated by its iBGP neighbor, no problem.
But what if, the route was redistributed from OSPF, to iBGP on router A, how does it update this route towards its iBGP neighbor router B? They share the same AS so by the route loop prevention rules, if router B isn't OSPF neighbor of router A, it will not get this route even if they're iBGP neighbors? yet, router A's eBGP neighbor, router C will get this route?
Thanks a lot for any answers.
