[Bio-Linux] Cloud computing...

[Tony Travis]

Martin Gollery wrote:
> Many people do not have the funds for a BlueGene/L, nor the need for
> such a thing on a regular basis, as WSU and VBI recently announced.
> The throughput was very impressive, but the cost is in the millions of
> dollars.

Hello, Martin.

AFIK, the 'cloud' is based on warehouses full of COTS servers, not on 
top500 HPC clusters. You're NOT going to get 'BlueGene' performance at 
knock-down prices. You get what you pay for.

> If a researcher had a job that was bigger than what he or she could
> easily do on the standard BioLinux system, then a 20 CPU instance can
> be started at Amazon for 80 cents (US) per hour. If there is a
> deadline looming, the data could be divided into, say, ten parts and
> run on ten instances for a total of 8 dollars per hour.

Iff you are only going to run your jobs once in a while, and Stewart has 
kindly set up a virtual image of Bio-Linux with everything you want to 
use already preconfigured, and all the databases you want to access 
transparently shared between all the anonymous 'cloud' nodes you get for 
8 dollars and hour, then you are in business.

> This does not 'lock you into a proprietary system'. Running HMMer on a
> cloud today does not prevent you from running it on your own server
> tomorrow, your mac the next day, and windows the next. Why would
> anybody think that? A Linux instance runs like Linux.

The 'cloud' itself is a proprietary system. It's the opposite of HPC 
clustering. What is the point of using virtualised Bio-Linux instances 
just to 'cluster' them?  If I want to use a top500 class computer, I'll 
join the queue on a real one. However, I follow Donald Becker's example, 
and use my own DIY Bio-Linux Beowulf cluster instead :-)

Bye,

	Tony.
-- 
Dr. A.J.Travis, University of Aberdeen, Rowett Institute of Nutrition
and Health, Greenburn Road, Bucksburn, Aberdeen AB21 9SB, Scotland, UK
tel +44(0)1224 712751, fax +44(0)1224 716687, http://www.rowett.ac.uk
mailto:a.travis at abdn.ac.uk, http://bioinformatics.rri.sari.ac.uk/~ajt