[179208] in North American Network Operators' Group
Re: BGP offloading (fixing legacy router BGP scalability issues)
daemon@ATHENA.MIT.EDU (William Herrin)
Thu Apr 2 19:21:37 2015
X-Original-To: nanog@nanog.org
X-Really-To: <nanog@nanog.org>
In-Reply-To: <CAKCUjRWVbLdBiqE__6F_EZ9-CSGhB8667kDDwY9gQN-oXcFKQA@mail.gmail.com>
From: William Herrin <bill@herrin.us>
Date: Thu, 2 Apr 2015 19:21:08 -0400
To: frederik@kriewitz.eu
Cc: "nanog@nanog.org" <nanog@nanog.org>
Errors-To: nanog-bounces@nanog.org
On Wed, Apr 1, 2015 at 1:01 PM, Frederik Kriewitz <frederik@kriewitz.eu> wrote:
> In
> practice the biggest problem with [Cisco 6500s] is their poor BGP scalability
> due to the CPU/memory limitations.
Hi Frederik:
Are you sure about that? I would expect you to hit the wall on the
TCAM long before CPU/RAM limitations. Check your log for errors along
the lines of "FIB_EXCEPTION: FIB TCAM exception, Some entries will be
software switched."
This might be masked as a CPU problem since routing for destinations
which fall out of the TCAM is forced up to the main CPU.
If this is your problem and you're talking about sup-720-3bxl's with a
1M TCAM, you can change how much of the TCAM is allocated to which
functions for now. Longer term, you either upgrade to something with a
bigger TCAM or you reduce the number of routes you carry by
introducing superset routes such as a default route and then filtering
out the more-specifics your customers are least likely to use.
Regards,
Bill Herrin
--
William Herrin ................ herrin@dirtside.com bill@herrin.us
Owner, Dirtside Systems ......... Web: <http://www.dirtside.com/>
