It's no secret that credit derivatives and exotic investments are all the rage on Wall Street. But as new alternative investment products are added to the mix, and portfolio managers and risk managers strive to get an accurate view of a fund's value and associated risk, systems are being stretched to the limit of their computing power. Fortunately, many firms began deploying grid architectures three or four years ago. Today, those grids have evolved into their second generations and are quite capable of handling Wall Street's ever-increasing demand for more and more computational power.
So it wasn't surprising when Chad Hersh, a senior analyst with Celent, wrote in a recent report, "Grid Computing: A Guide for Financial Services Firms," that "Grid computing is going to be a major investment for virtually every large company, financial services or otherwise." As far as prognostications go, Hersh's prediction looks like a good bet, in light of grid's obvious attractions.
In essence, a grid leverages all the different pieces of a firm's hardware -- which can include anything from a laptop to a mainframe -- to form a huge pool of computing power. When a task or calculation is submitted for processing, the grid software divides it up and farms those pieces out to wherever it has identified available computing capacity; once completed, the parts then are reaggregated. And because the grid can make use of idle desktops and inexpensive servers, for example, it provides firms with the ability to scale up their computing resources faster and more cheaply than previously possible. >>
For the securities industry, caught between an ever-increasing technology onus on one side and cost curtailment on the other, utilizing grid technology should be a no-brainer. Indeed, in Celent's report, which Hersh coauthored, the research and consulting firm predicts spending on grids by financial services firms will grow from an estimated $120 million in 2006 to more than $500 million by 2010. More tellingly, in the same time frame, the number of computations performed on grids are forecast to mushroom from less than 5 percent of the current number of computations performed by financial firms today to more than 30 percent.
Grids have proved particularly adept at processing computationally intensive tasks that traditionally take a lot of time and often run in batch-type cycles, according to Hersh. Examples cited in his report include derivative analytics, such as pricing, Monte Carlo simulations, portfolio performance analytics, and calculating the value and risk profile of trades.