The grid: The future of scientific networks.

February 5, 2002 6:01 AM

The grid: The future of scientific networks. "Built on the Internet and the World Wide Web, the Grid is a new class of infrastructure. By providing scalable, secure, high-performance mechanisms for discovering and negotiating access to remote resources, the Grid promises to make it possible for scientific collaborations to share resources on an unprecedented scale, and for geographically distributed groups to work together in ways that were previously impossible."
posted by talos (6 comments total)

Hope all the chattiness on the site lately isn't dissuading folks from posting more challenging items like this article. I'm looking forward to reading it. Thanks!
posted by sheauga at 7:45 AM on February 5, 2002

A very impressive article, my first reaction was "My god! They want to turn the entire planet into one giant supercomputer" Technically very cool, but I for one will be keeping a sharp eye on the. "Community authorization and policy" angle of this huge and very long-term project... "the policies that govern who can use which resources for what purpose" Distributed computing obviously has a big future, the question how control of this shared resource is handled is also big - especially when you take into account the human race's track record with shared resources...
posted by dorcas at 7:55 AM on February 5, 2002

ibm has made grid computing a big initiative along the lines of computation as utility.
posted by kliuless at 8:06 AM on February 5, 2002

Great article. After reading this, I'm encouraged by the advances that have been made. However, it is sobering to note that our need for increased computational power and bandwidth continues to outpace our ability to produce it, and probably always will. This is the perfect illustration of the old concept that for every question we answer, we create a staggering number of new ones. As for collaboration, I hope that the grid helps to open things up for more interdisciplinary research rather than traditional, compartmentalized approaches.
posted by gimli at 8:07 AM on February 5, 2002

Dorcas commented, but I for one will be keeping a sharp eye on the. "Community authorization and policy"

I work in the same division as Ian (author of the article above) and for a while I was an early "grid" user. Security and authentication have been a thorny issue for years. I still don't know if they have the right model. But there are some impressive things being done in the field.

One other major obstacle is administrative politics. Even when/if the security issues are settled there's still the problem of scheduling. When I want to submit a job to run at locations A, B and C, etc, there has to be some admin at each location to make sure the resources are available, that the machine doesn't hang, etc. Most of the time this means booting off local users (in the case of an all-important demo).

Now that IBM, Sun and a few other big corporate entities have taken notice it should be an interesting next couple of years.
posted by Qubit at 11:13 AM on February 5, 2002

However, it is sobering to note that our need for increased computational power and bandwidth continues to outpace our ability to produce it, and probably always will.

Conversly, but just as importantly, we're having one hellava time keeping up with our data -- yahoo was ok. hotbot (inktomi) was good. google is better. But we're drowning in seas of data.

From the article: CERN's Large Hadron Collider (LHC) among them, will produce multiple petabytes (1015 byte) of data per year.

Our real challenge is not more computations per second or more throughput in our pipes, but getting computers, and human-plus-computer systems to be better at understanding the data we produce.

Our tools for visualizing information are very impoverished, and, in the vein of the article, have *yet* to grow by an order of magnitude. We're still using OS metaphors designed in the 1960s (unix, et al) or 1970s (xerox park, et al).

We should go ahead and build grid computers, and yes, we'll need 'em, but that's not the problem regular guys like me face every day -- if there was a 1 GB or even 1 terabyte link between my office and my house, I could move mp3s back-n-forth pretty easily. But it doesn't help me view my collection, which now spans 3 different hard disks at 2 physical locations, (about 40 directories total -- perhaps 11GB).
posted by zpousman at 12:24 PM on February 5, 2002