March 19, 2007

Take a few minutes to step back from your day-to-day tasks and think about some of the cutting-edge technology that Wall Street uses almost as routinely as you use the telephone. If you think about the computerized trading and the ultrafast, high-performance processors that are vital to many aspects of business on the Street -- trading, risk analysis, market data -- it's astonishing.

For most of you, it's not news that financial firms are on the cutting edge of technology. You live it every day. Recently, though, a colleague of mine who is a long-time editor for InformationWeek, a CMP Technology property, and has an outstanding knowledge of business/technology trends, contacted me because he had trouble believing what he was hearing from a Wall Street CTO.

Essentially, the CTO told him that he didn't want to move his data center outside of Manhattan -- despite real estate and energy cost advantages -- because the few extra milliseconds was too much to spare for data to move to and from the data center. My colleague has a wealth of knowledge on technology and is very familiar with the efforts of Intel and AMD as they try to build faster and faster chips and processors, but to hear just how vital milliseconds (not seconds, or even hundredths of seconds) are to some firms took him a while to compute. Most professionals that either work in or follow the high-performance computing industry can recite stats about how many zillions of transactions a chip can complete in one second, or how much energy a chip uses, or how much cooling is required to prevent the processor from going supernova. But when the speed of a transaction is put into real business context, that is what grabs people's attention.

But the same reliance on bleeding-edge speed and technology can cost Wall Street, as well. Most recently, the NYSE's trading and order routing systems slowed to a crawl as they became overburdened on Feb. 27 with order flow. Adding to the problem, the computers that calculate the Dow Jones Industrial Average were so overwhelmed with data that they were calculating the Dow on a 70-minute delay. Once the the glitch was discovered, the Dow switched to a backup system that promptly digested all of the backlogged data at once. And once the data was processed, it appeared that the Dow precipitously dropped 240 points in 3 minutes, thus kicking in algorithmic trading systems that flooded the market with more buy and sell orders as the systems reacted to the wild swing in the DJIA. Oh, the pleasures and pains of cutting-edge technology.

ABOUT THE AUTHOR
Greg MacSweeney is editorial director of InformationWeek Financial Services, whose brands include Wall Street & Technology, Bank Systems & Technology, Advanced Trading, and Insurance & Technology.