E! Dives Into Online Data With Open Source

E! Networks, the producer of entertainment news for TV and the Internet, is assembling a low-cost data warehouse using software based on open-source database technology and inexpensive servers from Sun Microsystems.

E! Networks, formally known as E! Entertainment Television Inc., will use the system to understand what content the 6 million monthly visitors to its Web site are viewing.

Traditional data warehouses built using Oracle or IBM databases running on symmetric multiprocessing servers can cost as much as $1 million per trillion bytes of data they manage. E! Networks won't disclose the exact cost of its data warehouse, but technology VP Jeff Mayzurk says in an E-mail that the project costs "substantially less than more traditional solutions."

Dave Powell, CEO of database-management-software vendor Metapa Inc., which is providing the core software for the system, says the Metapa-Sun combination is priced at about $100,000 to $125,000 per terabyte. The data warehouse will have a starting capacity of 1 terabyte.

E! Networks has finished assembling the system and plans to start testing it this week. Once it's operational, company analysts will study clickstream data collected from E!'s Web site to understand what content visitors are looking for and what advertising is most effective. At the heart of the data warehouse is Metapa's Cluster DataBase software, which manages large volumes of data spread across clusters of inexpensive servers. It's based on the PostgreSQL open-source database. Mayzurk says he hopes the project demonstrates that data-warehouse technology is attainable by midsize companies with "conservative technology budgets."