Grid

re: Grid

Duncan Lamb — Thu, 20 May 2004 03:58:00 GMT

Grid computing = hype? "most if not all successful grid applications out there are the parallel processing applications of old?" This seems a bit too genralized to me... within Microsoft the corporation this may be true, but how about some real world examples?

Within a vendor-neutral definition of grid computing, there are many succeeful implementations. Does Google's architecture qualify? I think it would... SETI's @Home project? Ditto. Bittorrent? Gnutella? Spyware used on thousands of nodes to consolidate reports on marketing habits?

Of course, Microsoft, IBM, Oracle, et al will tailor a definition addressing their own strengths, and claim everything else is hype. The tools are there in alot of ways already, even if MS has not successfully developed a product to sell in that space yet...

My favorite sandbox in this area is ClusterKnoppix (http://bofh.be/clusterknoppix/) - boot it up off a cd (no installation required), set a few local machines to PXE/network boot (try your laptop), and watch a nice cluster form like magic with heterogeneous hardware. Try out a few things to experiment with, then when you're done reboot back to your regular machines (hard drives are never touched on any of the machines) until you're ready to take a look at it again. There is a whole community of people messing around in this space in the real world, although they themselves might not call it "grid," "on-demand" or something else.

Thanks for the summary by the way, it's always good to hear what's going on in these meetings and hear the perspectives :)

re: Grid

Darrell — Thu, 20 May 2004 04:25:00 GMT

Spyware doesn't count, since that is just an installed program sending in updates. That's not grid computing.

The @Home project is exactly what Michael said, traditional scientific applications. Mining data.

Google's architecture is more interesting.

Take Outs for 19 May 2004

Enjoy Every Sandwich — Thu, 20 May 2004 08:46:00 GMT

Take Outs for 19 May 2004

re: Grid

Michael Platt — Thu, 20 May 2004 14:35:00 GMT

SETI and @home are standard scienific apps.
Google is more like a server farm than grid

ClusterKnoppix demonstrates what I mean about Grid, there is some really interesting work around making a grid type platform / system going on (another is http://www.softricity.com/products/index.asp ) which are not grid but are very powerful.

By the way Microsoft says nothing in this space (to my knowledege) and these are my personal opinions, not Microsoft's.

re: Grid

Shirley — Sat, 22 May 2004 07:26:00 GMT

Interesting the 15year old 'parallel' discussion goes on !

re: Grid

Michael Platt — Sat, 22 May 2004 20:26:00 GMT

Yes, What Shirley would that be? Ha, thought so!!

re: Grid

Savas Parastastidis — Fri, 28 May 2004 06:24:00 GMT

Well said Michael.

We've been trying for a year now to demonstrate to the Grid community that it is not necessary to come up with a new messaging infrastructure or new frameworks to building Grid applicaitons. We only need to focus on the Grid-specific high-level services and let the industry deal with the WS "pluming". Unfortunately, the OGSI and WS-RF efforts demonstrate that there are people out there trying to offer infrastructure solutions before having identified problems or limitations with the current one. Not to mention the introduction of the completely wrong conceptual model, which is object- rather than service-oriented in nature.

http://www.neresc.ac.uk/ws-gaf for more info. More documents are comming up there.

.savas.

re: Grid

Michael Platt — Fri, 28 May 2004 14:33:00 GMT

Thanks. I do think that the concept of a grid platform is really important and that some of what the grid community is doing is very important. It's just they shouldnt replicate other work going on in this area.

re: Grid

Mon, 28 Jun 2004 18:53:00 GMT