[52234] in North American Network Operators' Group
Re: [nanog]software routers
daemon@ATHENA.MIT.EDU (Vadim Antonov)
Sun Sep 22 05:28:13 2002
Date: Sun, 22 Sep 2002 02:27:01 -0700 (PDT)
From: Vadim Antonov <avg@exigengroup.com>
To: Koji Hino <hino@ccrl.sj.nec.com>
Cc: <nanog@merit.edu>
In-Reply-To: <20020922.003343.132706659.hino@ccrl.sj.nec.com>
Errors-To: owner-nanog-outgoing@merit.edu
On Sun, 22 Sep 2002, Koji Hino wrote:
> From: Vadim Antonov <avg@exigengroup.com>
> Date: Sat, 21 Sep 2002 03:06:56 -0700 (PDT)
> :> The CPUs are quite faster nowadays, and you can get things like _quad_
> :> 300MHz PPC core on an FPGA plus 20+ 3.2Gbps serdes I/Os - all on one chip.
> :> So building multigigabit software router is a no-brainer.
>
> Usual PCs don't have such fancy hardware...
They have quite fancier stuff inside, actually, from a technology point of
view. Your run-of-the-mill Pentium IV requires a lot more advanced
technology for design and manufacturing than the "platform FPGA" I
referred to. The enormous design costs for those marvels of engineering
are only bearable because of high volume.
> If you call it software router, then all "network processor based
> router" should also be called software router, right?
"Software router" in my book is something which relies on programmable
general-purpose devices for implementing packet routing functions.
Network processors do not qualify (though they're nice, when they are
actually available, which is not often - because chip vendors tend to drop
niche products pronto when in a bind). One thing I learned well is to
keep exotic stuff out of designs, because it never seems to be available
w/o high-volume commitments, and even then tends to come a year later and
full of "design features".
--vadim