Analytics software company SAS has been working on (and has submitted a patent for) technology intended to handle risk analysis and compliance tasks in a high performance computing environment. The SAS technology uses a private server grid, a shared memory pool and parallel processing to allow high-speed calculations.
Large banks typically have farms of thousands of processors and hundreds of jobs, applications and processes that contend for those resources. SAS' technology pools the processors and RAM of a group of servers into a private grid, such that they're viewed as a single set of resources to the applications that need them. "This gives us far more memory to apply to complex analytical problems, says David Wallace, industry marketing manager, financial services at SAS, who spoke with Bank Systems & Technology in an interview last week. "When you apply that to complex risk calculations, where you're trying to value a large, complex portfolio across multiple states and time periods, the number of calculations that have to be done to calculate value at risk can turn into the billions. That takes traditional processing techniques a long period of time to solve, because some of that computation is sequential. We divide the total job of doing value at risk calculations for those portfolios into very small chunks across all the servers and memory, then sum it up on the fly."
Other companies, including IBM, Sybase, Gigaspaces and Violin, provide a way to do work in-memory, which can be much faster than going to an outside data source. SAS says it is not trying to get into the grid space. "It's built specifically for use by SAS applications," Wallace says. "We're using this new technology as way to further enable the vertical and horizontal solutions SAS has, in cases where the size of data sets or the complexity of the calculations is such that in a traditional environment, they're very long-running jobs. This takes that time down from hours or days to a few minutes.
Under the finance reform being finalized in Congress, banks will be required to step up their stress testing to better allocate capital. And regulatory requirements aside, "if banks want to truly optimize their capital and understand what they should do when a shock occurs, and what the impact of that shock is on a portfolio they're holding and the counterparts they're doing business with, those are questions that today are difficult to answer in the moment. We see the best practices will be to run stress tests on demand, that will augment management decision making in a way that provides a more precise view of what's going on across cash flows, exposure and volatility from a firm-wide standpoint instead of rolling up a series of siloed responses from the systems that exist today."
SAS is working with customers to fine-tune the product and has partnered with HP to develop it. The initial risk business areas SAS is targeting for the product are liquidity management, stress testing and scenario analysis and capital optimization.