Hi Everyone,
We are an eBGP multihomed to two transit providers (and a number of other peers on a peering exchange). Failover is pretty painless when one of the providers goes hard down (i.e. eBGP session drops).
However, if one of the providers has an issue within their network (such as flapping internal link, flapping iBGP sessions), then our edge router can not detect the failure and remove routes / advertisements to the bad peer.
What is the recommended approach to detecting a "soft" failure of a provider? The current solution is to manually alter the import/export policies on the problematic BGP session, but this is obviously not ideal as it requires manual intervention.
We're using Juniper MX at our edge. Should I be looking into use RPM probes to monitor each circuit from each Juniper MX at my edge?
No comments:
Post a Comment