Using Princeton Softech Inc.'s Ager 2000, IT managers took a snapshot of the bank's operational data--including loan schedules, mortgage payment schedules, and recent transactions--and aged it to 2000. The bank chose dates that were considered particularly problematic, such as Jan. 3 (the first business day of the new year), Feb. 29, and the last day of each month.

Caught In The Act
The testing revealed a few problems. For example, the bank needed to make corrections to an interface program developed in-house that passes transactional data between applications, says Terry Hughson, a programmer analyst in Old Kent's new product development operation. The program was still using a two-digit year format, and in 2000 would not have correctly translated data from one application to the next. Hughson says the problem was solved by making corrections to the program's code. Had they not been caught, Hughson says, the computer system might have rejected some transactions.

A primary benefit of using a data-aging tool such as TransCentury File Age is its ability to handle different data types, such as relational data and data in legacy flat files, says United Stationers' Niedzwiecki. Without such a tool, company programmers would have to write separate programs or utilities for testing. At the IBM credit union, Jorve found Data Commander to be a similar time saver by carrying out the data conversion automatically without additional programming. "It took us a month to do what would have taken us years to do manually," Jorve says.

In some cases, attempts to update business data could backfire. Gartner Group's Vecchio warns against the windowing approach many IT shops have taken. With windowing, code is added to applications that produce or use data to interpret whether a two-digit year begins with 19 or 20. For example, programmers add code to an application that instructs it to interpret all two-digit dates between 00 and 50 as beginning with 20 and all dates between 51 and 99 as beginning with 19.

Y2K experts say problems may crop up if the window is not big enough to cover all the dates within a company's data. That can happen if programmers developing the windowing logic are unfamiliar with the organization's data. For example, if a hospital has medical records from 1930 or a bank has loans that mature in 2060, the applications that use windowing could incorrectly interpret the dates. "Your windowing technique has to start at your oldest piece of data," says Kathy Tucker, Computer Associates' senior VP of application life-cycle solutions.

Testing the application and its windowing code against the data can help catch such problems. Being familiar with what dates a database contains can prevent them from occurring in the first place.

Analysts say problems can also occur if data transferred between departments or companies contains dates that were created using different windows. Cable service company Cox Communications, a division of Cox Enterprises Inc. in Atlanta, is using Data Commander to check data that is transferred between Cox and its financial institutions and suppliers. Cox's goal is to maintain date-format consistency within its databases of customer-billing information, service records, and other files, says Jim Cayo, application systems project manager.

It's not too late to address problematic data, and some testing tools can help. However, Meta Group analyst Shawn Bohner says companies shouldn't be completely reliant on such tools. IT departments should focus efforts on making contingency and disaster-recovery plans, he says. "Essentially, data corruption can be a time bomb waiting to go off."

return to page 1