IBM said the integration technology which will launch at the end of 2003, would also give end users real-time access to data beyond DB2 and rival databases. Using native XML capabilities in DB2, users will be able to access data from conventional sources, such as applications, as well as unstructured information stored in documents or e-mail messages.
The software will further IBM's approach to integration, which is designed to provide access to data without requiring users to move the information from its source and reformat it to be DB2-friendly.
The company has already used the federated approach in products such as its DiscoveryLink technology, which lets users in the life sciences industry touch multiple databases, applications and search engines with a single query.
IBM is the only large vendor that is pushing ahead extensively with this kind of virtual database concept, said Philip Russom, an analyst at Giga Information Group. One of the enablers for its approach is the fact that DB2 has the ability to index data stored in other repositories, he added.
IBM has yet to disclose the middleware platforms that will be supported by the new data integration software. Russom said the company is likely to use its own WebSphere integration and WebSphere MQ data messaging products.
The concept of federated databases has been around since the 1990s, but companies are only now starting to view it as desirable, Russom said. For one thing, extracting large amounts of heterogeneous data and formatting it for use in a database requires extensive network resources as well as a data warehouse, which can cost $1m (£620m) a year or more to maintain.
In theory, IBM's upcoming technology will allow users to avoid those costs while gaining real-time data access. The software will be released for beta-testing by June.