Subject: Re: Network data transfer speed (was: [fwd: NetBSD Live CD])
To: Ignatios Souvatzis <is@netbsd.org>
From: Greg 'groggy' Lehey <grog@NetBSD.org>
List: netbsd-advocacy
Date: 01/30/2006 16:16:57
--o/5eNASeIIpuMggS
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline

On Monday, 30 January 2006 at  6:40:55 +0100, Ignatios Souvatzis wrote:
> On Mon, Jan 30, 2006 at 12:58:41PM +1030, Greg 'groggy' Lehey wrote:
>
>> It looks like Linux is set to move ahead.  Last week, at the
>> Linux.conf.au in Dunedin, New Zealand, Van Jacobson presented a
> ...
>> Basically, his idea amounts to a complete redesign of the network
>> stack architecture, moving most of the code into userland and
>> eliminating the two-stage transfer to the user.  He showed results
>> that used only 20% as much processor time and ran twice as fast
>> (limited by the wire, not the software).  Very impressive.
>>
>> I asked him if he would be prepared to supply details to the BSD
>> projects, and yes, he is.  I'd suggest that this is a "must examine"
>> thing, and I'll post more as soon as it's available.
>
> This sounds a bit like a paper related to 4.4BSD about a redesign of the
> network stack. (Which, obviously, belongs to the design, not the
> implementation part...

Are you referring to a specific paper?  Van's slides were very clearly
aimed at the Linux 2.6 kernel.  It's very possible that they're based
on something older, of course.

Greg
--
See complete headers for address and phone numbers.

--o/5eNASeIIpuMggS
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (FreeBSD)

iD8DBQFD3ahRIubykFB6QiMRAlq6AKCRdPV2RY6lUGWH/zSrKfT9kE7/8wCdFChI
rW/K1by3XYeD6xn9ala82Oo=
=Hv7I
-----END PGP SIGNATURE-----

--o/5eNASeIIpuMggS--