[c-nsp] Big L3VPN on 6500

[Rodney Dunn]

On Thu, May 29, 2008 at 01:00:44PM +1200, Pshem Kowalczyk wrote:
> Hi All,
> 
> We provide L3 and L2 vpns services for our customers. Usually they
> have only a few routes (up to 1k), but this time it looks like we'll
> be providing transmission for a customer with almost 40k routes. They
> want to connect in 4 of our locations (using BGP). So far we ran the
> 6500s on standard cef settings:
> 
> FIB TCAM maximum routes :
> =======================
> Current :-
> -------
>  IPv4 + MPLS         - 512k (default)
>  IPv6 + IP Multicast - 256k (default)
> 
> 
> We have about 200k routes already (both ipv4 and vpnv4) and I started
> to wonder how the new customer will affect that. Are only routes
> (vpnv4 prefixes) that are in the fib counted (that's what I think) or
> will I see an increase of 4 * 40k entries (3 * vpnv4 + native
> peering)?

Only best path is loaded to TCAM.

 What's the easiest way to calculate the memory consumption
> that will be caused by the new customer?

Not an easy way unfortunately as the datastructures change a lot
so the best way is to baseline and graph memory in comparison to
number and type of routes.

Rodney

> 
> kind regards
> Pshem
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/