Redundant route reflector convergence time

[sumit menaria]

Hi,

I have two route relectors in our network as of the best practices for implementation of RRs.They are configured to be totally redundant to each other.

However I can see everytime one of the route reflector goes down ,the PE routers have a traffic disruption of 2-3 seconds.

In an ideal case ,if the route reflector goes down ,then the PE routers will store the route received from the secondary RR and since the next hop will remain the same as advertised in the prefixes by the primary RR ,there will be no change in the data path ,only at the control path there will be convergence.

Can anybody tell me why this 2-3 seconds disruption is taking place in case of failure of one of the route reflectors.

Thanks.

Regards

Sumit

[Vinit Jain]

Could you please share the configuration of the RR and the PE routers.

Also , please share the show ip bgp vpnv4 uni vrf <name> <prefix> and show ip bgp vpvn4 all summ output from the PE router.

PE's can take time to converge and based on that the CEF will be updated. I am not really sure how big is your vpnv4 table but it adds to the convergence delay.

In order to overcome traffic disruptions in SP environment, you can use features like BGP PIC and BGP AddPath.

Hope this helps

Regards

Vinit