/. MS vs. university Beowulfs
Doug J Nordwall
nordwall at pnl.gov
Fri Feb 22 09:45:04 PST 2002
I guess my view is pretty simple, and the same way that MS rules the
desktop...Linux on clusters works, and works well. Why should I bother
switching? The applications that run on clusters generally just would
not be the sort of thing that I would normally think of on windows, and
even if it could benefit from it (say with a gui) it could just as
easily be a front end.
I asked our windows guys (a dozen or so seasoned vets) about automatic
installation, and they said it's in there, but requires third party
apps. Now, I found it hard to believe, so I'll look for someone to shoot
it down, but if that's the case, then there really isn't much reason :)
The other thing that was a good point on slashdot was the cost per node.
DO I have to license a copy of windows for every machine in the cluster?
When you see people building clusters out of $250 machines (really, I
know a guy who did), a $200 OS or $100 OS, or a $50 OS is a big deal.
Especially when it's x16, or x32, or x128.
On Fri, 2002-02-22 at 08:18, graham.mullier at syngenta.com wrote:
> Well, you could have a look at http://www.windowsclusters.org/ as pointed
> out by one of the slashdotters.
> http://www.windowsclusters.org/projects.htm looks the most useful from this
> site, if you're trying to find some real info.
> Personally I'm not interested in a comparison of a full cluster of linux
> nodes vs one of W2K nodes, but of what use can be made, in a corporate
> environment, of heterogeneous clusters, pulling in and using desktop compute
> cycles, a la seti at home or whatever. Obviously it's a completely different
> game, and my main question is whether the return is worth the effort (i.e.
> can you get enough unused compute cycles out of desktop machines to make it
> worthwhile). The chapter on Condor in "Beowulf cluster computing with Linux"
> (ed. Sterling; 2001) implies that it can be worth it.
> Any views?
> Graham Mullier
> Chemoinformatics Team Leader,
> Chemistry Design Group,
> Syngenta, Bracknell, RG42 6EY, UK.
> direct line: +44 (0) 1344 414163
> mailto:Graham.Mullier at syngenta.com
> -----Original Message-----
> From: Velocet [mailto:math at velocet.ca]
> Sent: Fri 22 February 2002 15:34
> Have fun wading through the blind linux advocacy and silly jokes to try
> and find any content.
> If someone can summarize any real information they grok from this massive
> thread (ie how good is computational clustering on M$), please post it here.
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
Douglas J Nordwall http://rex.nmhu.edu/~musashi
System Administrator Pacific Northwest National Labs
More information about the Beowulf