I have a design question. We have a Telco that is providing us with a primary(200Mb) and backup(50Mb) MPLS circuits. These circuits will terminate on a 2911 each. We run EIGRP and the Telco will be using eBGP. We re-distribute the BGP into the EIGRP.
core <==> 2911 <==> Telco device <==> MPLS cloud
Question - What is the best way for us to failover to the backup when the primary fails?
We have thought of a couple of ways to do this but I want see what the community thinks.
Enable the (BFD) feature , bidirectional forwarding detection for BGP. It has significant convergence improvement on the BGP WAN.
Here is the link:
Add the following command to the interface connected to 200Mb link:
bfd interval 50 min_rx 50 multiplier 3
under BGP configurations add the following command:
neighbor x.x.x.x fall-over bfd
where x.x.x.x is BGP neighbor router.
However you need to do this on ISP router side as well.
are you terminating each MPLS link to a separate 2911 router ?
assuming 2 diffrent routers:
- Redistribute the BGP routes of the second telco with high EIGRP metrics ( assuming you receive same routes from both Telecos ) and the failover will be delayed based on BGP timers once the primary BGP peer gose down and the route disappeared from the BGP routing table then it will stop being redistributed into EIGRP from the primary router, you can tune BGP timers for quicker convergence time
- you may summarize each link routes in to one summary route if its summariseable using static route, then associate this route with and IPSLA track that track the reachablity to the BGP next hop and redistribute this route into EIGRP instead of redistributing BGP, you need a careful testing to this option and this concept to be use don both ends and if a more specific route route disappeared from the BGP traffic might be blockwholed
- you may use IPSLA with EEM but this will make it a bit complicated,
- using BFD to bring down the bgp quicker need both end to use BFD in this case your end and the telco must use it !!
- using HSRP/VRRP with tracking NOT an option for you because you are already using EIGRP which is quicker and the routers are no a demarcation point between L2 and L3 in your case
hope this help