Beneath Arthur Ashe Stadium, sealed under soaring concrete stands and guarded by a small army of US Open personnel, Kathleen McGraw is guiding me through a series of bunker-like war rooms, where ThinkPads outnumber human beings. As McGraw, an IBM employee, points out various servers and the engineers manning them, I find myself wondering just how many computers it takes to run a tennis match. I count two players, two rackets and a ball. Isn’t all this hardware just overkill?
According to IBM, that’s the point. In the world of computer hardware, overkill is just another word, like redundancy or mirrored processing (I guess that’s two words), for covering your ass. While hardware failures aren’t good for any business, they’re especially painful for large media portals hosting major sporting events with a raft of big-name sponsors. Hence, it should be no surprise that the US Open Web site is backed up in three locations; one crashes, the other two step in. But what fascinated me wasn’t the back-up plan. It was the $10,000 server doing credit analysis and folding proteins in the corner.
When traffic peaks at the US Open Web site (say, during finals matches), the server must have the capacity to juggle hundreds of thousands of requests. But when traffic is low (say, what, 3 AM?) the server has a lot of “white space,” better known as nothing to do.
When IBM’s i5 server isn’t serving up scores from the Open, the company has it crunching consumer credit data (albeit, a demo in this case) and “calculating algorithms of proteins in various states at the molecular level to understand how they fold” for something called the Blue Gene Science Program. Suffice to say, the little i5 is hard at work. Shades of the busybody manager at my first restaurant job (“If you don’t have work to do, I’ll find some for you!”)
What other overlooked resources would make good targets for helping to increase productivity and efficiency? A couple years ago, I seem to recall an organization tapping into unused home PCs during the night to borrow processing power (think Seti@Home and similar efforts). Any others come to mind?