07-08-2009 04:16 PM - edited 03-04-2019 05:22 AM
I have two routers and two MPLS WAN links at all locations running EIGRP with a single AS to both WAN providers and when a link failure occurs on one of the WAN links traffic would failover to the other link within seconds. And now recently I moved one of the WAN links at all locations to a new MPLS WAN provider and I am running eBGP with this new WAN provider, so today on one router I am running both EIGRP and eBGP and EIGRP only on the other router and on the router that runs both EIGRP/BGP I am redistributing BGP into EIGRP and under BGP I am using network statements advertising the local networks to my BGP peer and all seems to be working fine except when WAN link on the router that runs EIGRP only fails, it takes a good 90 seconds for traffic to failover where as before when I have both routers running EIGRP only to both WAN providers failover was a lot faster within 30 seconds versus now 90 seconds more less.
Can someone tell me if this is a normal behavior ?
Thank you in advance !! I appreciate any inputs/suggestions !!!!
Danny
07-09-2009 05:18 AM
Danny-
BGP was designed to support large, scalable networks and provide it stability at the cost of rapid failover. EIGRP was designed for rapid failover/link detection. There are ways to speed the BGP convergence, google it or here's a doc that I found.
Hope that helps.
07-09-2009 08:07 AM
Hello Clark,
Thanks Clark for responding !! I understand what you're saying but it takes about 90 seconds for the network to converge vs lest than 30 seconds with EIGRP. Is it typical for BGP to take that long for it to converge ? I've never work with BGP before and would like to find out if that's a normal behavior.
Thanks again !!!
Danny
07-09-2009 10:33 AM
IF EIGRP took 30 seconds (that seems long to me), BGP at 90 seconds is reasonable.
07-09-2009 10:40 AM
EIGRP actually took about 10 seconds to converge (I was a bit off).
Thanks Clark !!! I appreciate your inputs !!
Danny
07-09-2009 12:28 PM
As Collin notes, BGP wasn't really designed for fast convergence, like many of the IGP protocols. Further, Cisco defaults, I believe, have been configured to deal best with full Internet BGP route tables, and can often be safely adjusted to be much quicker when working with, the often found, many, many fewer (vs. Internet) BGP routes within MPLS VPN.
Here's a Blog that discusses this: http://networkers-online.com/blog/2008/12/bgp-performance-tunning-convergence-stability-scalability-and-nsf-part-1/
Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: