Data warehouses accumulate terabytes of data for deep analysis. That's one way to create business intelligence; Truviso offers another. Its software performs continuous, real-time analysis on incoming data. Truviso got its start selling to financial firms. It views retail, RFID, IT infrastructure, and Web operations as being ripe for its approach, too. ----John Foley

Truviso's adaptive query processor makes it possible to run dozens, even hundreds, of SQL queries against data as it comes in from a variety of sources. Truviso's algorithms can comb through historical data, but the real advantage is on-the-fly analysis of stock price, sales, or clickstream data, so actionable information can be revealed immediately.

Truviso's architecture consists of three subsystems. A core engine, based on the open source PostgresSQL relational database, manages queries and other data management tasks, while an integration framework provides connectors to data feeds and source systems. The third subsystem, with graphing and charting templates, is used to create browser-based front ends to view query results. Truviso segments data by time or number of records, then runs queries against those data sets.

Financial firms are early adopters of complex event processing, and Truviso is going after that market with an algorithmic trading version of its platform. Logistics, inventory, network monitoring, and RFID networks are other environments where the company sees potential for up-to-the-minute business intelligence. Competitors include Coral8, Progress Software, and StreamBase Systems. Truviso has struck go-to-market partnerships with GoldenGate Software, Hotspot FX, and Reuters.

Franklin is a computer science professor at UC Berkeley, and Krishnamurthy has a Ph.D. from the school. Pearce was previously with ANTs Software, Hitachi Data Systems, Pyramid Technology, and Teradata.