[106303] in North American Network Operators' Group
Re: Software router state of the art
daemon@ATHENA.MIT.EDU (Florian Weimer)
Sat Jul 26 06:41:47 2008
From: Florian Weimer <fw@deneb.enyo.de>
To: Adrian Chadd <adrian@creative.net.au>
Date: Sat, 26 Jul 2008 12:41:29 +0200
In-Reply-To: <20080723161740.GK11922@skywalker.creative.net.au> (Adrian
Chadd's message of "Thu, 24 Jul 2008 00:17:40 +0800")
Cc: nanog@merit.edu, zzuser@yahoo.com
Errors-To: nanog-bounces@nanog.org
* Adrian Chadd:
> 1 mil pps has been broken that way, but it uses lots of cores to get there.
> (8, I think?)
Was this with one packet flow, or with millions of them?
Traditionally, software routing performance on hosts systems has been
optimized for few and rather long flows.
Anyway, with multi-core, you don't need funky algorithms for incremental
FIB updates anymore (if you don't need sub-second convergence and stuff
like that). As a result, you can use really dumb multi-way trees for
which a lookup takes something like 100 CPU cycles (significantly less
for non-DoS traffic with higher locality).
