Have financial services firms captured the Holy Grail of technology? That's what some observers say as firms implement grid computing-an architecture that helps harness all the computer power across an enterprise. For Schwab's David Dibble, it's been a long time coming. Too long.
Imagine being able to access the power of every computer server in your company and turn it into a virtual supercomputer accessible by every segment of your bank-whether it's in Bangkok or Baltimore. Say hello to grid computing, the new "It" concept sweeping the industry that aims to squeeze the most power out of technology infrastructures. All the big financial firms are moving to implement their own grid-computing systems, but community and small banks are still woefully behind, say analysts.
The lure is seductive: Firms can expect to trim hardware, maintenance and labor costs by up to 25 percent by shifting hardware, storage and software configurations as needed. But it'll cost big bucks to prepare: Industry consultants estimate the financial services marketplace will spend $4.89 billion by 2008 on grid computing components. Although the technology is not new-it's been in use for years at computation-intensive industries like engineering and biotechnololgy, it's certainly new to the financial services industry.
Grid computing is the pooling of computer power across an enterprise into a flexible architecture that allows banks to shift resources where and when they're needed most. "The concept of providing computing power as a utility is extremely attractive to end-users requiring faster transactional processing and scenario modeling, or to IT staff looking to control costs and reduce complexity," says Celent Communications senior analyst Michael Haney in a research note. "The ability to have a cluster, an entire data center, or even resources, spread across the Internet itself (and) function as a single system that can be managed as a service, rather than as individual components, will increase business agility, reduce complexity, streamline management and lower costs."
Firms are probably using only 35 percent to 40 percent of all their computer power across their company, observes Dushyant Shahrawat, a senior analyst in the securities & capital markets practice at TowerGroup in Needham, MA. "Grid computing links everything so companies can share (computer) resources across silos to do those heavy, intensive calculations often needed in finance. Instead of buying new hardware and software, why not use what you have right now?"
Grid computing aims to accelerate the development and performance of on-line, real-time, computing- and data-intensive applications, thereby managing the underlying IT infrastructure to help banks and brokerages make business-critical decisions faster and more accurately.
"Grid computing is just beginning to be hot. We are in the early stages, but the ramp is going to be steep," says Benny Souder, vp of distributed database development at Oracle Corp., who calls himself "Mr. Grid." Oracle is a provider of grid computing software.
Grid computing is so hot that the version of Oracle's flagship database software, released in September, is called Database 10g to underscore the firm's emphasis on grid computing. "The reason it's so compelling is all businesses have this universal experience: I'm trying to close the books and my computers and resources are overloaded, melting down," says Souder. "My IT department says, 'That's it. That's as much as we can do. We have to cross our fingers and hope.' Meanwhile, I've got hundreds of computers sitting idle. That breeds lots of frustration."
Shahrawat says grid computing is an especially useful tool for financial services firms, renown for their many independently managed silos. "This is an industry were there are barriers," he says. "Computers are often not efficiently utilized."
Large firms are jumping in because they have so much more to gain, Shahrawat says, but the price of enterprise-wide implementation is not cheap. "The cost involved is that it's not economic for a firm to do it as a narrow base," he says. "The investment to implement only makes sense across an enterprise, like a full-blown implementation across a large brokerage firm, which could cost millions of dollars. But to get the full value, it needs to be done enterprise-wide."
But the payoffs are huge. "People aren't really aware of how much it can provide," Shahrawat says. "In the financial services industry, market changes are measured in seconds and delays can mean the difference between profit and loss," says Michael Sharma, vp of sales for financial services at Platform Computing, the largest player in the financial services sector. "(Grid computing) can remove technology constraints and drive optimal performance."
The decade-old Toronto-based firm offers three grid-computing products-LSF, Symphony and Intelligence-which are being used by 1,600 clients around the world, including Groupe Societe Generale, Royal Bank of Canada, Landesbank Baden-W?rttemberg, Chase Manhattan and World Bank of Canada, Capital One, Deutsche Bank, JPMorgan Chase and RBC Financial Group.
"Deploying grids is not just about workload," says Sharma. "It's also about measuring workload and identifying the key applications and deploying them when needed. What are the right applications to put on the grid? ...If you're JPMorgan Chase and you have 4,000 machines and, at any given time in the market, you can take those 4,000 machines and turn their attention to the most mission-critical issue at the moment. That's the secret to our software. ...It's logical, but this industry has always sold individual silo-based applications and hardware. It's the spaghetti issue." The firm claims to have the only grid-computing products that support all commonly adopted standards, including .NET and J2EE.
Among the financial services firms making inroads using grid technology-using it for number-crunching jobs in risk management, analytics and portfolio rebalancing-include Bank One, JPMorgan Chase, Deutsche Bank, Merrill Lynch and Charles Schwab one of the firms would talk about their projects, however, except Schwab-so it must be doing something right.
"The grid becomes part of the (technology) infrastructure," notes David Dibble, evp of Charles Schwab Technology Services, who says the firm expects to implement its new grid system, a joint project built with IBM, by the fourth quarter. "The days of one application, one box, is gone, absolutely gone. The grid is the key enabler. ... The greatest area is the opportunity to develop and deploy far more sophisticated client-oriented solutions that we couldn't have done this fast before. It gets the creative juices going."
The San Francisco firm's financial advisers had been struggling for a year with how to more efficiently process the multitudes of Monte Carlo simulations for client portfolios, which requires heavy duty computing power. "These are high-performance simulations," says Schwab's Dibble, who notes that calculations that used to take eight minutes on its Forecaster portfolio-management product now only take 30 seconds. We couldn't afford to build the biggest supercomputer environment," says Dibble. "We're not a national lab. We're trying to make money on this."
Grid computing made sense for a firm that is conscious about cost and sensitive about providing high-quality service. "No matter how you slice it, most of our machines are idle most of the time," says Dibble. "The vast majority of machines in most sectors are used less than 50 percent of the time."
Dibble declined to guess how much money the firm is shaving off its technology expenses with a grid system. "The savings come in cost avoidance," he says. "As you better utilize computer resources, those who pay for them originally now have someone else to share the burden of cost. And you're also avoiding having to buy a dedicated infrastructure."
Souder points out that grid computing brings so many more benefits than cost savings. "It helps you align your priorities with the business, and as your priorities change, they stay aligned," he says. "That is something that provides value for the entire enterprise." Moreover, says Don Russo, vp of financial services for Oracle, grid computing also provides another layer of security, since data ca be warehoused in fewer locations that can be guarded more thoroughly.
Like all major tech implementations, standards are important, which explains why Platform is delighted that many major system vendors, including Dell, IBM, HP, NEC and SGI, have adopted its grid technology to support their enterprise computing strategies. In September, the firm created three global acceleration centers-in New York City, San Jose, CA, and Washington, DC-to help organizations prepare their applications and infrastructure for grid computing. The centers will provide technical consulting, application integration and professional services, in partnership with key system vendors, application software vendors and professional services organizations. In the next few weeks, Platform is expected to announce new OGSI-compliant products, key partnerships and application integrations, which will be offered through the three centers. The firm declined to provide prices.
But Oracle says firms want simplicity and that its shrink-wrapped software, which is standardized, will be an affordable off-the-shelf solution for many firms. Officials declined to provide prices. In addition to releasing Database 10g, reportedly the first database with advanced self-management, clustering and storage management for enterprise grid computing, the firm also released Application Server 10g. This application allows firms to run existing applications and Web Services on enterprise grids unchanged; also launched was the complementary Enterprise Manager 10g, which monitors and manages entire Oracle grid infrastructures-from databases and applications to storage-from a single console. At the heart of the product are the Real Application Clusters, which officials say is the only clustered database software that runs packaged applications unchanged as additional servers and storage are added. The clusters, which have been in development for more than a decade, are being used by more than 600 customers in 44 countries.
Shahrawat says widespread application of grid computing in financial services is still "three or four years" away. "Since it requires collaboration across technology providers, unless those providers are on the same grid computing map, it will take time. ...It's too early to be either cynical or optimistic," he says.
IDC analyst Mark Melenovsky expects grid computing to be one of the key IT expenditures for financial services over the next five years, particularly if belt-tightening continues to dominate the sector. "Grid computing is a very big deal," he says. Consultants note that many retail banks will end up using that extra computing power for data-warehousing tasks; transaction processing; pattern-detecting for anti-money laundering; risk modeling in such knowledge-based systems like Bayesian belief networks and causal networks; portfolio optimization, balancing and performance analytics; derivatives pricing; trading analytics, credit-default analysis; and mean-variance forecasting.
"Despite all the potential advantages of grid computing, it should be recognized that its use in the financial services industry is still in its infancy," says Haney in his research note.
Russo agrees, but points out that there is simply no turning back now. "Grid computing is the next new wave that is going to become a standard across this industry for managing computer-intensive resources," he says. "It's really one of the Holy Grails we've been going after as an industry ever since the client server first peeked over the horizon in the late 1980s. This gives us the opportunity to take advantage of the distributed-server technology that launched back in the late 1980s with some of the control and value that we got out of the glass houses of the '70s. It's the best of both worlds."
