Grid

[btiffin]

Old guy ramble;

Get on the GRID kids. This is big. The web lets us access information. The GRID will let you access a computer built out of millions. Make business plans, find a science niche, ... and get ready for processing power beyond belief. Your bigger jobs may queue, but the power potential of this is big.

If you ponder what it would be like to be Bill back in the 80's, Yahoo, ... Google just a few scant years ago, well in my humble opinion that opportunity is upon us again. This is still new, and fast movement now could be lucrative. At worst, you learn something. At best, you own the top dog and the envy.

Just sayin'. Go GRID. I'm serious. Having lived through pre PCs, 150 BAUD modems, to GOPHER to http to grid ... grid is BIG.

End ramble.

Start questions.

Anyone here in a school that is part of a VO? (Virtual Organization with GRID security tickets). Or know anything about the LCG VDT? (Large Computer Grid, Virtual Data Toolkit). I want to add and integrate OpenCOBOL to the mix. Who do I talk to?

Cheers

[Zeroth]

Tsk, tsk, tsk, Btiffin. I thought you were more pragmatic than that. My issues with this "GRID" or the "cloud" as it is called, is that conceptually, it is very different from reality. Mostly cloud computing is a buzzword. Essentially the only service that satisfies properly the potential of cloud computing is the Amazon Web Services.

Heres the cons(and I'll list which don't apply to Amazon):

-less control of the platform
-constrained to what OS/tools they offer(Not Amazon)
-legal issues concerning data storage, retention, encryption, government access, etc. (Not Amazon)
-bandwidth issues
-typically, cloud computers are, to remain cost effective, the extra servers a company owns to cover peak surges(Very definitely amazon)
-reliability

Pros:
Lots of computing power
Scales at command

Note, however, that cloud computing does NOT solve your problems for you. And some problems are not inherently amenable to grid/cloud computing. It has to be lots of small but cpu intensive jobs, or a few large, but easily split jobs. Google's MapReduce is one that takes care of several issues, reliability, bandwidth, etc. If its a problem that could be solved by MapReduce, then its one that works well on grid/cloud computing.

Personally, I would not rely solely on cloud computing, but a mixed infrastructure. The grid/cloud gives me the opportunity to scale as needed(only Amazon offers this so far, with complete control of the OS and installed tools), while, I'd like to, most of the time, keep the data under my control. During a peak, sure, lets offload the load. But otherwise, to minimize chances of legal issues, lets keep it here.

[btiffin]

Hmmm, I respect that opinion. But I believe different. The Grid as CERN envisions as I see it will allow for large scale very fast analysis of huge datasets. For institutions at first, us soon. This changes the game a little; common folk getting advice on the stock or futures markets, someone writing a global anti-spam sweep, doctors cross referencing billions of samples, it's big. I'll be humble and say imho, but not really. It's big. Really.

A simple idea right now could make someone the inventor of the next big thing, a big thing that is coming. I can't tell you what it could possibly be, but I expect to hear about it soon.

Cheers