Every day, the Bank of New York gives its institutional-investor customers an investment-portfolio update, including a snapshot of the day's transactions. And each day, the bank stores these reports, along with copies of monthly and quarterly summary reports, which also are sent to the pension-fund operators, investment firms, insurance companies, thrifts, and government agencies that make up the bank's clients.

But by last year, the bank's database was close to capacity. Those client reports, though rarely accessed, must be maintained for years to meet Internal Revenue Service regulations. The growing volume of stored data threatened the system's performance and the bank's plans to grow its institutional-investor business.

One solution was to expand the system with more hardware and database storage capacity. But that would be costly and would only solve the problem until new capacity limits were reached. "If we wanted to continue to grow this database and continue to serve the number of clients we expect to have, we needed to archive this data offline," says Peter Keaveney, VP and head of the bank's Investment Accounting Systems Division.

Last fall, the bank installed Active Archive software from Princeton Softech Inc. to offload older data from the bank's operational IBM DB2 database. The system archives data on optical disks and maintains its referential integrity. Although the data is archived offline, it can easily be accessed with a browser-based application. Data also can be brought back to the operational database in bulk or selectively. The project cost about $1 million, Keaveney says.

Today the bank's operational database holds daily client-portfolio reports going back 105 days and two years' of monthly, quarterly, and annual reports. Older reports are stored in the Princeton Softech data archives, which now hold 1.2 billion rows of data.

The bank saves between $50,000 and $60,000 a month by not having to buy additional hardware to accommodate the rapidly expanding operational database, Keaveney says, and it's cut CPU use by 20%. He says database reorganizations and backup-and-restore tasks also are quicker.