LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

LVS talk at UKUUG Linux 2001 conference

To: lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Subject: LVS talk at UKUUG Linux 2001 conference
From: Martin Hamilton <martin@xxxxxxxxxxxxx>
Date: Thu, 05 Apr 2001 17:18:16 +0100
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Content-Type: text/plain; charset=us-ascii

A few weeks ago I noticed the call for papers for this, i.e.

  http://www.ukuug.org/events/linux2001/CFP.shtml

and absent mindedly wrote in offering to do a talk on LVS if there was
a) some interest in this on the part of the organisers, and b) nobody 
else proposing to do such a thing.

So, I just got an invite from them to come and do this.  However...

Given the calibre of the usual speakers at these conferences (e.g.
Alan Cox, Stephen Tweedie, Miguel de Icaza, Hans Reiser, ...) it would
probably be a much more worthwhile session if we could get Wensong
himself or one of the more prolific contributors to the LVS code to
come along...  I understand that some help from UKUUG and their
sponsors may be available towards travelling expenses.

If any of you guys is interested in doing this, let me and
linux2001@xxxxxxxxx know - I'm happy to bow out in your favour :-)

Cheers,

Martin

PS The info about the JANET Web Cache Service on the LVS deployment
page is now a little out of date...  FYI:

On our busiest day to date (6th April) we shipped 118m proxy HTTP URL
requests or 1.2TB of Web 'content' to downstream institutions
connected to the JANET network.  We're now regularly getting over 100m
URL requests in a single day and shipping over 1TB of content.  This
is with three server farms (at Leeds, London and Manchester), each
with its own director.  Each director currently uses IP tunnelling and
the WLC scheduler to talk to a number of back end Linux boxes which
are running the Squid proxy cache server on top of ReiserFS.

Each of our cache server farms currently has a 100Mbit/s uplink to the
core of the JANET network, but plans are afoot to give us either a
separate 100Mbit/s feed for each cache server or a 1Gbit/s feed for a
cluster as a whole.  At the moment we're bandwidth constrained rather
than constrained by server capacity, so we should see soon whether a
(non-NAT!) director is capable of scheduling over 100Mbit/s of
production traffic :-)

More at http://wwwcache.ja.net/ for anyone who's interested...




-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.4 (GNU/Linux)
Comment: Exmh version 2.1.1 10/15/1999 + martin

iD8DBQE6zJrIVw+hz3xBJfQRAtIYAKCuoimHfN68BnjAmEjVUGUdm92HmgCfaF4F
rpBgemC1vLsfX3fWnROSqXo=
=+al7
-----END PGP SIGNATURE-----



<Prev in Thread] Current Thread [Next in Thread>
  • LVS talk at UKUUG Linux 2001 conference, Martin Hamilton <=