Re:Distributed computing? (Score 1)

There are a few reasons:

1. The granularity of the tasks that are farmed out to SETI@home screensavers (and other similar projects) are relatively small. They are not particularly large in terms of memory or compute requirements.
2. The input and the output of these types of tasks is small. The data movement requirements are not all that large.
3. These tasks only talk to the central server. The communication pattern is a very simple star network. Sure, it requires some interesting provisioning at the central server(s), but these are well-known client/server problems.

The goal of the Grid is to turn up the knob on all of these - to deliver complex communication patterns, large grain computation, with large data sets, over a wide area, straddling multiple management domains. Applications that make use of this kind of infrastructure are interesting.

There a good number of these efforts in the U.S. and Europe, but very few in Canada. I wish Jonathan all the success in the world with this. I hope that this will help get Canadian granting agencies to notice how things could work. (Here's an exercise: Ask your favourite Canadian granting agency, say http://www.nserc.ca/, what their policy is on Grid computing.)

See the Grid Canada site (http://www.gridcanada.ca/) for information on a parallel effort to get resources to participate in a cross-Canada Grid project.

Darcy