A Little Confused
By templedf on Mar 29, 2007
In a comment to a previous post, someone said that "The day your business needs grid computing, you are already dead." That's such a wonderfully absurd statement, I just had to bite. In fact, I'd say the person got it exactly backwards. The day your business needs grid computing, is the day you know you're successful.
Grid computing is about scalability. Sun makes the world's biggest and baddest SMP machines, but even with the biggest, there are situations were big just isn't big enough. Grid computing gives you the ability to harness an entire network of machines to get the work done.
A good example is credit card fraud detection. Every single credit card transaction that happens in the world has to be processed and validated in an very short time window. That's a lot of volume with a very low latency: more than any single machine could ever hope to handle. That's why many banks are turning to grid computing. The same story applies to equities trading companies.
Another good example is the EDA market. I'm willing to wager that there's is not a single chip designer out there today that does not employ a grid during the design process. Car makers love grids, because grids enable them to do faster, more thorough crash simulations. Biomedical research groups love grids, because grids let them do complex analyses faster. Oil and gas companies love grids, because grids enable them to get more detailed models of underground wells in less time.
Contrary to the slightly confused comment, grid computing in the enterprise is a sign of life and prosperity. Dead companies don't use grids; they use Irix.