iBGP Failover

[benolyndav]

Hi

I recently configured iBGP between a 3750 stack and a pair of N5K's  where I brought up  x2 L3 connections from 3750 stack 

to N5K pair, over these L3 links I have a seperate instance of iBGP from 3750 stack to N5k's Both iBGP sessions are established and I do see x2 routes in the BGP Table and just x1 route in the RIB which I would expect, my issue is when I test a failover scenario I loose approx 12 pings before the second route is installed in the RIB and the switch resumes the forwarding of the pings, any idea how I can speed up this process,??

Core switch svi used for BGP

N5k's x2  Loopbacks used for BGP peerings

FYI this is a lab enviroment testing before rolling out to physical devices in production

[Elliot Dierksen · Accepted solution]

If you want faster convergence, that is the domain of an IGP like OSPF or EIGRP. For an EGP like BGP, stability is more important than rapid convergence.