[Bio-Linux] Cloud computing
Tony Travis
rwt017 at abdn.ac.uk
Wed Dec 3 10:01:34 EST 2008
Dan Swan wrote:
> [...]
> I think what has happened is people have seen how horrendous the Grid
> situation has become, and how far it has deviated from its own 'plug
> in and play' mantra for access to computing resources, mired as it is
> in certificates, middleware and authentication hell - and the cloud
> has been borne from this. It's a streamlined approach that offers
> most of what people want without all the horribly complexity.
Hello, Dan.
My experience with GRID is limited, but I agree with what you say. In
fact, I attended a Globus level-2 system admin course given by NeSC in
Edinburgh a few years ago, and decided on the spot not to use it, but I
relented, and went to the myGrid 'users' workshop. This was great from
the point of view of seeing what you can do with Taverna, but it did
nothing to improve my view of GRID middleware. Neither did seeing how
BLAST can be deployed on the GRID. I don't think GRID does well at a
scale most biologists normally work at. Many bioinformatics resources
that biologists do use are either web pages they browse, or web services
they consume, offered at no cost by large public-funded organisations.
My own interest is in heterogenous self-organising networks of p2p
computers. My original idea for 'biobuntu' was to use the Kerrighed
kernel, and do GRID communication in the kernel with XtreemOS:
http://www.xtreemos.eu/
This is the opposite of the 'Cloud', in which we would 'volunteer' our
spare resources to the network when we don't need them locally, but use
other people's resources when we have a big job to do. This works well
for loosely coupled problems, for example folding at home, and I'm not
alone in thinking that p2p is still under utilised in bioinformatics.
The XtreemOS project is intended to support virtual organisations, and I
think it is worth considering as an option for Bio-Linux development.
> People whose requirements outstrip the cloud I suspect will be better
> off served by existing Grid based provision.
This drives me crazy, because a lot of what biologists do is not 'BIG'
computing and an HEP billion-dollar solution is not appropriate for all
problems. I do realise that some problems of interest to biologists
require huge computing resources to be thrown at them, but supporting
what biologists actually want to do in the way that NEBC has done so far
is probably of greater interest and value to most biologists than the
actual hardware it runs on. In that respect, I think the issue of cloud
computing vs. rolling your own workstation/server/cluster/GRID is less
important than the continuing support and development of Bio-Linux :-)
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
More information about the Bio-linux-list
mailing list