IBM to Extend DB2 Integration

'Masala' to search heterogeneous apps.

IBM is working on new technology that extends the reach of its DB2 Information Integrator software to let users dig out data from every corner of an enterprise, whether in a data warehouse or outside of it.

The technology, code-named Masala, is designed to push the limits of DB2 Information Integrator, according to Nelson Mattos, an IBM Distinguished Engineer and director of information integration. One of its central functions will be to reach beyond DB2 Information Integrators current grasp of databases and stretch up to application packages from vendors such as SAP AG, Siebel Systems Inc. and PeopleSoft Inc.

Masala will build on DB2 Information Integrator by offering data replication thats several orders of magnitude faster than the speed of the current version, Mattos said. It will offer the ability to use free-form text to search an information infrastructure to get hits on relevant data using content-ranking capabilities.

Masala represents the next wave of information integration technologies to come from the Armonk, N.Y., company. It is part of a push to make IBMs DB2 database a better platform from which to integrate data, no matter where it is stored—in applications, data warehouses or databases from virtually any vendor.

According to Mattos, IBM is working on this next stage of information integration to answer the pleas of customers who say that, after they deploy information integration infrastructures, they still need to drag in data that lies beyond traditional data repositories, including e-mail or repositories of Adobe Systems Inc.s PDF files.

IBM has long championed the data management concept of federation, or leaving data in heterogeneous databases and applications, on different platforms, while abstracting the data for delivery to an application.

"Federation is a key component of information integration," said Mattos. "Without it, you have nothing."

Federation is central to information integration because without it, applications must know where data is stored. Data has to move from one place to another—for example, from a Sybase Inc. database to an Oracle Corp. database. The journey entails coding against different APIs, which all have to be ported, and this creates extra work with which costs are associated, Mattos said.

Federation brings a level of abstraction that eliminates the dependency between the application and the physical location of the data. Rather than incurring the extra work of porting APIs, an application using federation can issue a request to DB2 Information Integrator, which knows in real time where data is.

Finally, Masala will bring additional self-tuning capabilities to Information Integrator. Administrators will spend less time configuring, deploying and monitoring infrastructures, Mattos said.

Masala will ship to a limited set of beta customers this year. The first commercial products will go out next year.