On Monday, February 10, 2014 09:01:35 PM Spyros Kakaroukas wrote: > It’s going to take a bit more labbing before we decide > whether we want to actually implement this or not, but > it seemed like an interesting idea so I thought I’d > share. > > Feedback would be appreciated!
I recommend this approach. I was working with SPAG and they had some really good test results of running BGP-SD on the ME3800X. There shouldn't be much difference in expectations for the ME3600X (I think they both have 2GB RAM). So yes, use BGP-SD to receive the full feed, but keep it away from the FIB. Forward these on to downstream customers to handle the control plane. Have 0/0 and ::/0 programmed into the FIB to handle the data plane. The only think to look out for is CPU utilization during route churn and link flaps, as well as growth of the BGP tables and its effect on RAM. Keeping an eye on software and how it's optimized for BGP route processing is also something to consider as you upgrade/downgrade software during the lifetime of the system. Not having to run eBGP Multi-Hop for customers that require a full BGP feed is a huge win. Mark.
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
