On 2011-06-01 21:09, Eric Morin wrote:
Hi
I have an application where I have a 7206 with NPE-G2 (1G RAM) that currently has a full table from an eBGP peer and a full table from a co-located IBGP peer. I want to mesh this guy to two other IBGP peers that will also send their full tables. There is roughly 400Mbps (adding both direction) with an OSPF process (<30 routes) and an ISIS process (<20 routes).
First of all - cisco-nsp@ is a better place to ask such question and if you search archives of it, you'll find a lot of comments for such config.
I haven't been very successful at finding real life BGP scaling for this platform and was hoping that someone out there may be doing something similar (4 x full table feeds with ~400Mbps) and can provide feedback on performance and/or stability. I have soft reconfig inbound enabled on the current two feeds, I assume that this will make a difference in this application with regards to available memory for all 4 feeds?
With 1GB of RAM you may run short (check current actual usage), but I'd recommend in such situation to rely rather on the route refresh standard capability, than the soft reconfig - it eats memory per-peer, and with route refresh you can get almost the same, trading bandwidth on the link for RAM in the RP. NPE-G2 can take 2GB of RAM, so if that's going to stay I'd also recommend upgrading if you're running short of memory either way. Or upgrade to ASR 1k and move into the world of hardware-forwarding platforms. -- "There's no sense in being precise when | Łukasz Bromirski you don't know what you're talking | jid:lbromirski@jabber.org about." John von Neumann | http://lukasz.bromirski.net