Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

New Member

Multi-home with same provider, BGP convergence issues


I have sort of an interesting problem, that probably has an easy fix. I've ran into a continuous BGP convergence problem on the cores of the network I administer every time I bring up both of our Savvis connections (we have a single OC-12 and 2 1GigE connections in a multi-hop configuration).

The connections are on separate border devices separated again by a set of cores routers. There is a set of core routers per building and each building has its own Savvis connection along with other connections from other providers. All the equipment is Cisco gear. We've been running with multiple connections from Savvis for some time, but recently reconfigured the GigE lines to multi-hop to help with balancing, only then, did we experience this BGP convergence issue.

>From what my colleagues and I have read so far is that turning on "bgp deterministic-med" on all the BGP speaking devices should help in this situation, but I don't see how using MEDS is going to help us.

I can provide a .png of our current setup for reference or any further information needed. Any help anyone can provide will be greatly appreciated.


Re: Multi-home with same provider, BGP convergence issues

Enabling bgp deterministic-med removes any temporal dependency of MED-based best path decisions. It ensures that an accurate MED comparison is made across all routes received from the same autonomous system (AS).

It will not alter the convergence time.

Based on the number of prefixes you are receiving, you can tune the timers (scan time, advertisement interval etc) for faster convergence


New Member

Re: Multi-home with same provider, BGP convergence issues

The problem we are seeing is that the network never fully converges.

The RIB continuously changes who is the best route for some, but not all networks.

From the research I've done so far, there are a few things that I could also try.

1. Strip the MEDs and turn on deterministic-med

2. Remove 'redistribute static' and 'redistribute connected' from the IGP and set the next hop self command on the neighbour statement.

Any thoughts?