[177534] in North American Network Operators' Group
Re: scaling linux-based router hardware recommendations
daemon@ATHENA.MIT.EDU (Ken Chase)
Tue Jan 27 00:07:47 2015
X-Original-To: nanog@nanog.org
Date: Mon, 26 Jan 2015 23:52:05 -0500
From: Ken Chase <math@sizone.org>
To: nanog@nanog.org
In-Reply-To: <BEC5618A-198B-4B7E-B38E-10392C0B63BC@neilson.net.nz>
Errors-To: nanog-bounces@nanog.org
On Tue, Jan 27, 2015 at 04:59:12PM +1300, Alexander Neilson said:
>For the CCR1036-12G-4S with one full table, one domestic table (NZ - ~26k entries) some peering and iBGP full convergence took about three minutes forty seconds last time I timed it from cold.
That's terrible.
I dont know what model that is or appropriate deploys but I think a couple of my peers
use these and report similar times for their models for 500k+ routes. Still too slow.
I think the single threaded nature of the routing table manip is at fault with
the 36-but-slow cores (mikrotic). Im not sure how you get around this without
drastically rewriting the kernel, which puts you out on your own developing
new fundamental tech.
I'd be more comfortable with full-cpu models (like xeon based for eg.)
> From: Faisal Imtiaz <faisal@snappytelecom.net>
> Under 30sec (more like 15 to 20) on an i7 based Mikrotik for full BGP Tables.
Ya, that.
/kc
--
Ken Chase - math@sizone.org Toronto
