IBM spices up corporate search

The company said DB2 Information Integrator, code-named Masala, will allow businesspeople to query multiple data sources on company networks, including text documents and e-mails, relational databases and multimedia files. Using the software, a customer service agent could quickly pull up customer information stored in different applications.

IBM sells DB2 Information Integrator as an add-on to its DB2 relational database for querying disparate data sources. Masala, named after an Indian spice mixture, will introduce a text-based search engine based on research from IBM's WebFoundation project.

The combination of text search and the traditional database-querying tools will enable businesses to search both "structured data sources," such as database records of transactions, and unstructured information like file systems, according to IBM. The company has been using the text-based search in its internal portal, which encompasses about 10,000 Web sites.

The company said Masala differs from traditional Web-based search technology because it is tuned for data other than just Web pages, using algorithms to search for content not interlinked as Web pages typically are.

As it moves into corporate search, IBM will compete with search companies, including Google, which sells a search appliance for company intranets, as well as other database software companies.

One of the "pillars" of the upcoming Longhorn version of Windows is WinFS, a file system designed to simplify search desktop PCs and company networks. Microsoft has delayed planned delivery of WinFS until 2007.

Oracle, IBM's other large database rival, is also investing in software to handle types of data besides the numerical information usually stored in relational databases. The company is expected to announce a new content management product, called Tsunami, later this year.

IBM charges $5,000 per server processor for Masala and $15,000 for each data source connector.