A new computing revolution is dramatically changing how Financial Services (FS) institutions use their IT resources to drive competitive advantage. Whether it's called on demand, grid computing, adaptive enterprise or service-oriented architecture, this is not just marketing talk. It is estimated that grid spending in financial services will jump from $59 million in 2003 to $680 million by 2008, according to the Tabb Group. IDC says grid computing technology will generate $12 billion by 2007.
Why grid? Ever-increasing demands for application performance, growing computing needs, and evolving regulations and budgetary pressures are imposing considerable business and technology challenges for financial services line-of-business (LOB) executives. Traditionally, computing power is confined to specific business lines, which may or may not use all computing capacity. Grid changes that by dramatically increasing CPU utilization within the business unit in addition to removing IT barriers, allowing FS firms to share their computing resources more effectively across multiple lines of business. This ultimately provides the organization with noticeable flexibility and efficiency improvements in IT management.
Today, many leading financial organizations are using commercial grid computing technology to accelerate application speed and create virtualized environments where risk management, credit analysis or homegrown applications can run as needed, faster and more efficiently than ever before. A recent survey of key IT decision makers from global organizations shows that grid computing is escalating as an IT investment priority in 2004, particularly among large commercial organizations, according to Platform Computing. Indeed, early adopters in capital markets, insurance and retail banking, such as JPMorgan Chase, RBC Financial, France's Groupe Societe Generale, and Capital One, are already on the road to grid and are documenting benefits.
For example, JP Morgan Chase credits grid with savings in excess of 20 percent on total compute cycle costs, according to ComputerWorld. They were also able to build a new credit-trading application on top of a grid infrastructure in 10 weeks rather than in the normal five-month timeframe. Similarly, Groupe Societe Generale is building a Windows-based grid to allow their commodities traders to do large scale, intra-day scenario analysis. Pre-grid, they could not have even conceived of this.
It may seem like grid technology is a huge undertaking but, in fact, the road to grid simply entails a well-researched business case, a proper map and a first simple step: think big, start small.
1. Plan your destination
The first step in a journey is determining where you want to go. In grid terms, this means clearly identifying the opportunities available through grid technologies. "Think big, start small, and grow incrementally" is a pragmatic approach to grid. It also means identifying a specific set of initial applications that will benefit most from grid technology and showing return on investment to justify the level of resource.
Early grid adopters are finding that massive increases in available compute capacity allows them to run more simulations, more frequently with more fidelity. But while leaders in adoption have implemented grid enterprise-wide, some institutions have leveraged the benefits of grid computing by choosing a single application in a specific LOB. For example, one global financial services firm chose grid computing to provide on-demand access to computing power for end users. Its vision accelerated the performance of applications and delivered calculations and reports faster to business analysts, who could quickly act on them.
2. Pave the road for future trips
Once an organization has determined which of its applications or LOBs will be grid-enabled, there are two important elements that must not be overlooked: efficient deployment and cultural obstacles.
There is seldom just one application that can be grid-enabled within a large company. In many cases, hundreds or thousands of applications are good candidates. So while a small deployment is a great first step, over time, grid computing can be extended to service multiple applications, in multiple lines of business, and multiple geographical locations. Therefore, to guarantee that grid is implemented quickly and efficiently in future deployments, it is critical to create a grid-enablement "assembly line" or repeatable process to handle the volume. Best practices to keep in mind include leveraging software engineering disciplines to establish robust processes and ensuring that common templates and components are used throughout all enabling activities.
It is also essential to recognize the obstacles to grid adoption that culture and firm politics can create and to devise a strategy for overcoming those barriers. Some business units may fear that they are losing control of their resources or sacrificing flexibility. When JPMorgan Chase set out to have seven separate financial risk management systems share computing power, the company realized that "pooling all kinds of IT resources from all areas of the company" required facilitating communication in open forums. Whatever the methodology, any grid computing project requires a highly coordinated approach among parts of the organization that traditionally may not converse frequently.
3. Grid is an evolutionary path
Like any newly cut road, the path takes a while to get worn and can be bumpy at first. This path to grid is evolutionary; it is not hard to envision that grid will re-define back-end computing within FS organizations the same way that the Internet and Web-based applications have revolutionized front-end computing. Grid software vendors continue to work closely with key FS application vendors to grid-enable commercial risk management applications, and the pre-integrated solutions they offer make it fast, easy and safe for FS organizations to move off-the-shelf applications into a shared grid infrastructure.
Robert Boettcher is vp of financial services for Platform Computing.
