Recently I realized I had got a bit fuzzy on the difference between clustered computing and grid computing. I found a good explanation at a page entitled Enterprise Grid Computing Solutions. According to them, my conflation of the two is…

As far as I know, you’re right about Windows–no booting without a video card. But, to tell you the truth, I’m not sure how much the cost of a crappy, built-into-the-motherboard video chip adds to the cost of a motherboard. My guess is pennies. Sure, 100,000 * any number of pennies is a significant figure, but it’s probably very insignificant compared to the cost of the rest of the hardware and the facility.

I completely agree with your sentiment that “Perhaps Microsoft can leverage their existing infrastructure and prove that GUI tools can do everything the UNIX ones can and more, but they are starting with less experience.” They’re running behind on this one so far. You’re also right that our (Digipede’s) product is *not* aimed at solving that problem–we’re aimed at hundreds of machines in the enterprise, not tens of thousands across the world.

You may be interested to know that most of our customers use our product on headless server nodes.

One last thing: I’m also interested in programming models, and you’re right that many applications just can’t benefit from grid. But we’ve worked hard to make a programming model that anyone who is adept at OOP strategies can employ–no knowledge of crazy threading principles or arcane MPI terminology required!

I, too, am interested to see how the whole thing works out. For quite a while now I’ve wondered about the Sun and Google partnership that got a bunch of publicity last year. Sun knows all about selling computes by the hour; Google has some of the world’s largest data centers, and MapReduce to boot. Think they’d ever open that up ala Sun Grid?