nodalpoint.org - grid - Comments

What need is there?

Duncan — Tue, 30 May 2006 03:14:37 -0400

The word GRID reminds me of the word GENE. It is either used in a vague and ambiguous way or it has multiple but precise definitions that are completely different. Despite this I think both words are useful and will continue to play an important role in the life sciences :)

If you consider any kind of

ejain — Sat, 27 May 2006 09:08:26 -0400

If you consider any kind of distributed computing or clustering to be grid computing, then I guess you can indeed say that grid computing "plays an important role in the life sciences". Of course, if that's all it is, what need is there for the term "grid"? Wasn't the original idea of the grid concept to create a standard mechanism (not tied to any specific application) for consuming (or even sharing) resources such as processing power and disk space, similar to the way the power grid works?

Grids, Google etc.

Neil — Sat, 27 May 2006 01:52:49 -0400

It's been interesting to watch the minor 'Google backlash' of recent months. This seems to happen to all companies that start small, promote themselves as more ethical than others then achieve success. I remember a similar media reaction to the Body Shop back in the 80s. Google's foray into the Chinese market seems to have caused a lot of this reaction.

I've always had my doubts about grid computing and bioinformatics. As the original post points out, there's always been a lot of esoteric debate about "what is a grid" rather than "what can I do with one". In their cluster computing presentations, Bioteam, for whom I have a lot of time, have been quite dismissive of the grid concept in the past, simply because noone can quite define what it is.

Keeping it simple, let's say a cluster is a pile of machines in one physical location, tightly-coupled through a network switch and dedicated to a few specific tasks. A grid is a bunch of machines in numerous locations (perhaps scattered worldwide) that in theory can cooperate on any task with appropriate tools and services to couple them together. In the former case your tool is very clearly defined, in the second it is not. Which is why currently, local HPC resources for typical applications such as BLAST have the advantage. I simply can't imagine posting off my sequences to a Google-like service and getting back results in any reasonable time with the current state of grids.

Google @ help

Animesh — Fri, 26 May 2006 09:12:50 -0400

Guys I read about GDisk thing [ http://blogs.zdnet.com/Google/?p=121 ] sometime back and then I also heard about google providing the ssh access to their supercomputer[ http://www.vmunix.com/mark/blog/archives/2004/07/27/what-is-google-build... ]. It may all be rumor but if it is going to be true, then I feel this will really be like a Venter to NIH thing looking at the above post!

______________________"The Answer Lies in Genome"______________________
http://fuzzylife.org/