[Beowulf] [owner-chemistry at ccl.net: CCL: Linux Cluster Building Workshops]
eugen at leitl.org
Fri Dec 9 05:54:20 PST 2005
----- Forwarded message from "Kenneth Geisshirt kenneth],[geisshirt.dk" <owner-chemistry at ccl.net> -----
From: "Kenneth Geisshirt kenneth],[geisshirt.dk" <owner-chemistry at ccl.net>
Date: Fri, 9 Dec 2005 08:27:58 -0500
To: "Leitl, Eugen -id#3h6-" <eugen at leitl.org>
Subject: CCL: Linux Cluster Building Workshops
Reply-To: CCL Subscribers <chemistry at ccl.net>
Sent to CCL by: Kenneth Geisshirt [kenneth=geisshirt.dk]
Andrew D. Fant fant(~)pobox.com wrote:
> O'Reilly, whose titles are
> usually the last work for IT topics like this, has published 2 disappointing
> volumes for this as well. The first was simply Building Linux Clusters and has
> been withdrawn from the market. The follow-on "Building High-Performance Linux
> Clusters" is better, but still is less that one might help.
I will recommend "Building Clustered Linux Systems" by Robert W. Lucke.
> If you simply want to put together 4-8 PCs to run a single application for you
> or a single research group, I would suggest looking at Rocks (
> http://www.rocksclusters.org) as a simple way to get something up and running.
The Rocks distribution is easy to get going. And it's a Red Hat based
distro, many commercial software packages will be easy to install.
Moreover, the many 3rd party package repositories (atrpms.net etc.) can
be integrated and you will have access to a large code base.
> My biggest bit of advice, however, comes from a bumper sticker that the
> electrician's union hands out around here. "Wiring Is Not A Hobby". If you are
> going to be running multiple applications from multiple sources/vendors, going
> above about 10 systems, having users from more than one fairly tight-knit and
> cooperative research group, or making the results from this cluster the basis
> of your research plans, you probably want to get an IT professional involved.
I can second that (I'm one of those IT professionals). Linux clusters
ain't rocket science but the complexity can be huge. If you start from
scratch you have to decide which Ethernet switch is best, how to wire
the stuff (low level) and which policy to implement in your batch system
(high level). If you build a cluster every second year it's almost
impossible to keep up with the development in this area. An IT
professional might build 5-6 clusters every year so pick his brain!
-= This is automatically added to each message by the mailing script =-
To recover the email address of the author of the message, please change
the strange characters on the top line to the @ sign. You can also
look up the X-Original-From: line in the mail header.
E-mail to subscribers: CHEMISTRY at ccl.net or use:
E-mail to administrators: CHEMISTRY-REQUEST at ccl.net or use
Before posting, check wait time at: http://www.ccl.net
Job advertisements: http://www.ccl.net/jobs
Search Messages: http://www.ccl.net/htdig (login: ccl, Password: search)
If your mail bounces from CCL with 5.7.1 error, check:
----- End forwarded message -----
Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org
ICBM: 48.07100, 11.36820 http://www.ativel.com
8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 189 bytes
Desc: Digital signature
More information about the Beowulf