[c-nsp] Making SUP720 cope better under BGP load

[Blake Dunlap]

Honestly I'd do your BGP peering with another platform at the scale you're
at. Since you're talking IXP do you really need forwarding plane and
control plane to match by letting the 65 do the customer peering?


On Fri, Dec 7, 2012 at 10:32 AM, Chris Evans <chrisccnpspam2 at gmail.com>wrote:

> In the past my company has ran into these issues. We helped it some by
> doing a hold-queue of 4096 on the interfaces and enabling jumbo frames
> where possible.
>
> It sounds like you're just running into a CPU issue though, which is one
> reason we moved away from the 6500/7600 platforms for this use case. Very
> very slow convergence due to slow CPU.
>
> On Fri, Dec 7, 2012 at 10:07 AM, Simon Lockhart <simon at slimey.org> wrote:
>
> > On Fri Dec 07, 2012 at 09:54:08AM -0500, Randy wrote:
> > > Have you considered a CoPP policy to limit the rate of BGP convergence?
> > > Not sure if it would help with so many peers but it might lessen the
> > > pain on your 3 full tables.
> >
> > No - I'm not doing any CoPP at the moment - but probably should.
> >
> > Are there any cookbooks / cribsheets for using CoPP to rate limit BGP?
> >
> > Simon
> > _______________________________________________
> > 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/
> >
> _______________________________________________
> 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/
>