Some companies are addressing only the programming logic of applications for Y2K, and that may not be enough, say analysts. Applications might contain problematic data, such as dates recorded as two-digit years. That could lead to computer errors resulting in botched production schedules, incorrect customer bills, or other mishaps. Compared with efforts to update applications, "not as much attention has been paid to verifying that data is correct," Gartner Group analyst Dale Vecchio says. "There is the potential for a big problem."

Analysts say some companies are overlooking the issue because they view Y2K as a programming problem rather than a data issue. In such cases, Y2K projects are often headed by programmers, rather than IT operations personnel.

Packaged testing tools are available that can help find data problems. United Stationers Inc., an office supplies distributor, is using TransCentury File Age software from Platinum Technology Inc. (recently acquired by Computer Associates) to test data in its mainframe order-entry and billing applications. "We're testing every app, front-to-back," says Bob Niedzwiecki, manager of application development systems and Y2K projects at the Des Plaines, Ill., company. The testing hasn't uncovered any major date-processing problems, but will continue for the next few months.

However, such tools have uncovered problematic data for some companies. Late last year, IBM's Mid-America Employee Federal Credit Union discovered a number of errors, such as 30-year mortgages that indicated a maturation date in 50 years. Those errors were not likely to cause computer failures, says David Jorve, senior VP of IS. But years from now, the IT staff could have found itself solving issues that should have been addressed before 2000. "They wouldn't have stopped the show," says Jorve, "but there would have been problems."

The credit union used Data Commander from Blackstone and Cullen Inc. to "age" some 50 Gbytes of data. Data-aging tools roll forward all the dates in a data file to reflect how they would be interpreted at a specified date beyond 2000. Applications are then run using that aged data to look for potential problems.

Jorve and his staff aged its 1998 data to 2026, because the monthly calendar was the same and provided a more consistent testing environment. Jorve focused his search on files that used two-digit date fields, as well as date fields that used a format that was different from applications: day/month/year rather than month/day/year, for example.

Data-aging tools can also help with the testing of remediated applications. Aging data to a post-2000 date and running an application against it provides a more realistic test than running the program against simulated data, say analysts and users.

Old Kent Financial Corp., a retail banking and financial-services company in Grand Rapids, Mich., aged data to test homegrown and packaged applications. "If you don't age the data, you're not testing the program logic correctly," says Jeff Jones, senior corporate technology VP at Old Kent.

continued...page 2

Photo by Jodi O'Shaugnessy