Integrating All That Data

This was just our year to attack the problem," says Christian Hudson, CIO with Swiss American, a custodian bank operating under the Credit Suisse umbrella. The problem he speaks of is none other than the feared obstacle of data integration.

Usually, the need for data integration arises as a result of business silos springing up around a financial institution. Silos are created when, in a mad dash to get a new product to market, a distinct and separate technology infrastructure is quickly thrown together. Once a few of these silos invade a business, data integration becomes an all-consuming and costly endeavor. The alternative, however, is a flood of trade errors and fails, leading to an exception-processing nightmare.

"There are valid reasons for allowing a silo to come into existence, and technology should not be an impediment to business," says Hudson, who has been with Swiss American for about two years. "That said, you can't try to half way integrate it and hope everything will work out. You either have to leave it completely independent and deal with the costs - both monetary and processing - or there has to be an integration plan back into your data-integration model," he asserts.

Hudson says that's why Swiss American, which has $70 billion under custody, is currently working to bring its silos back into the fold. "We have silos on the data side. We have three pricing feeds, two corporate-actions feeds and a number of other things that generate our symbology [security] master," he says.

There are two reasons this situation is untenable, says Hudson. The first is cost. "None of those feeds are free ... There are cost savings associated with potentially being able to get rid of a vendor." He notes that the savings potential is around $100,000. As part of an effort to reduce the number of vendor feeds coming into Swiss American, Hudson and his staff have just completed a re-evaluation of the bank's current providers. He declines to name any of his data vendors.

The second danger of processing trades in a silo-riddled firm, Hudson says, is the confusion created with trying to perform data rationalization. "It's like having people in your firm speaking different languages while trying to communicate about the same transaction," he says. "If you don't have that data integrated, if you are not rationalizing that data, eventually your firm is going to have problems."

Having just completed the data-vendor review process, Hudson says that his next step is reviewing recommendations from his staff regarding which feeds should stay and which should be shut off. He will also evaluate suggestions as to how the firm can move away from the silos currently in existence.

"We are going to create a single source and force everyone to program against it," Hudson says. The question, he adds, is, "How are we going to manage the technology integration that's going to have to happen in order to move the programs and applications that reference this data from the silos into a common database?" He continues, "What we will need to do is go either silo by silo or en masse and say, 'OK, I am going to move you off of this data in your own silo, and now you are going to come to the center.'"

Hudson explains that Swiss American is essentially trying to move from a silo-heavy environment, or decentralized environment, to a centralized, integrated, hub-and-spoke architecture. "Instead of everyone doing everything on their own, we need to work with the business units and the applications that run in those units to reference off the center spoke instead of off the satellites," says Hudson.

The whole data-integration project, he says, is about reducing errors. "You are basically attacking the exceptions-handling process, which is created by mismatches in data due to silos having different data. But instead of trying to attack it at the message level, you are attacking it at the foundation level."

The hardest part of the project, he cautions, is not integrating the data and collapsing the silos into a central database, but rather making sure that once silo-free, the firm stays that way. "It's easy for the organization to work very hard and centralize, integrate and rationalize the data against multiple feeds. The hard part is not letting another silo crop up." Anthony O'Donnell has covered technology in the insurance industry since 2000, when he joined the editorial staff of Insurance & Technology. As an editor and reporter for I&T and the InformationWeek Financial Services of TechWeb he has written on all areas of information ... View Full Bio