[157849] in North American Network Operators' Group
Re: Whats so difficult about ISSU
daemon@ATHENA.MIT.EDU (Mikael Abrahamsson)
Mon Nov 12 02:30:23 2012
Date: Mon, 12 Nov 2012 08:29:59 +0100 (CET)
From: Mikael Abrahamsson <swmike@swm.pp.se>
To: Bryan Fields <Bryan@bryanfields.net>
In-Reply-To: <50A099CC.4050106@bryanfields.net>
Cc: NANOG list <nanog@nanog.org>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
On Mon, 12 Nov 2012, Bryan Fields wrote:
> And they only have to process maybe 2mbit/s of control traffic during
> busy hour. The rest is handled by dedicated hardware/ASIC's. Each one
> has a fully redundant hardware circuit pack and a bunch of monitoring to
> switch over in case one fails.
I'd imagine it's also because some are written in a language especially
designed for the task.
http://en.wikipedia.org/wiki/Erlang_(programming_language)
"... It supports hot swapping, so that code can be changed without
stopping a system.[2]"
I've been told some people are doing routing control plane implementations
in erlang just because of these features, but I'd imagine there is a
hurdle getting enough programmers who are experienced in the language.
--
Mikael Abrahamsson email: swmike@swm.pp.se
