All this is stretching firms’ technological resources. Few can afford massive supercomputers to quickly perform the enormous number-crunching tasks, and IT departments are often hard-pressed to get overnight batch runs completed on time, never mind attempt intra-day calculations.
But while their centralised computing resources are struggling to cope, firms often have many microprocessors standing idle at any one time. And global firms will have offices full of unmanned machines on the far side of the world at any point in the working day.
So what if there was a piece of software that could sneak around the network and scavenge their spare processor capacity, putting it to work on the big analytical computations?
New York-based software start-up DataSynapse’s LiveCluster is just such a product. It is already running in the Charlotte, Virginia, offices of Wachovia Securities (formerly First Union Capital Markets), where it has led to up to 90% improvements in analytical performance for the firm’s interest rate derivatives group. The firm is extending the technology to turbo-charge the analytics of its foreign exchange exotics, currency swaps and mortgage-backed securities businesses. DataSynapse says it is also in production or active trials at nine other firms.
LiveCluster uses peer-to-peer (P2P) computing, where a number of machines on a network talk directly with one another, sharing resources and passing files between them. It can have many applications.
Some of the technologically advanced banks have been experimenting with the concept for a while. JP Morgan, for example, has created its own P2P network at its London interest rate derivatives group, where it claims it has achieved a 700% improvement in the cost-efficiency (a different measure from that used by Wachovia) of the group’s 100 or so machines (Risk June 2001, page 66).
DataSynapse has tailored P2P for the financial markets and turned it into a relatively easy-to-implement product that can help firms quickly harness their idle computer power for demanding pricing and risk calculations. It has also addressed a number of issues, such as security and reliability, which arise from distributing large amounts of sensitive information over a network.
LiveCluster creates a virtual supercomputer from the idle processing power in an organisation. “It provides a single image of the resources,” says Peter Lee, co-founder and chief executive officer of the company, and manages the allocation of tasks to individual machines and moving tasks on when a user begins working on their machine once more.
The latest version offers the ability to distribute very large datasets across DataSynapse’s processing environment. This is particularly useful where firms have very large portfolios and want to perform a number of analytical processes on them, such as sensitivity, VAR and profit and loss calculations, says Lee.
The implementation at Wachovia has been so successful that users want more of their analytical processes brought into the P2P environment, and more traders and risk managers want access to the technology, says Joe Belciglio, managing director, trading technology at the firm. “We’re able to do things during the day that previously we could only do overnight,” he says.
Prior to purchasing LiveCluster, Wachovia considered developing its own P2P application, but its policy is not to develop its own infrastructure technology if it can avoid it. The technology can also help with disaster recovery preparations – a more pressing requirement in the wake of September 11, says Belciglio. “The analysis can run anywhere, so if one or more machines fail, the software can restart that portion of the work on other machines,” he says.
Other firms using DataSynapse’s product include leading Wall Street and European financial institutions and energy trading companies, says Lee. In one instance, a firm has raised the number of pre-deal simulations it does per trade from 4,000 to 50,000. One energy company demonstrated the technology to its oil exploration unit that is now running trials with LiveCluster for analysing its huge seismic data files. Only Wachovia is willing to talk openly about its implementation for now, although other firms confirm off the record that they are impressed with the company’s technology.
Although DataSynapse is a small, new company, with around 30 employees, it is backed by $20 million in funding that it has raised over the past year from a group of venture capital firms. Michael Krupka, managing director of Bain Capital Ventures, which led DataSynapse’s recent round of financing, says it looks for “innovative technologies, solving large problems, backed by strong management teams with domain expertise”, and that DataSynapse met these criteria.
LiveCluster meets Risk’s criteria of a new, innovative product, soundly implemented, proven in production and with wide potential benefits for derivatives trading and risk management, and hence wins our technology product of the year award.
The week on Risk.net, December 2–8, 2017Receive this by email