Data analytics startup lands MySpace

MySpace is analyzing massive amounts of data generated from its Web site using software developed by a start-up.

Aster Data Systems of Redwood City, Calif., was founded by graduates of the Stanford University computer science Ph.D program and is due to come out of stealth mode on Tuesday. The company's data analytics software, called nCluster, is meant to be an alternative to database management systems such as Oracle's and the IBM DB2 system. While challenging two industry giants is a tall task, Aster's first customer announcement is an impressive one.

MySpace, according to Aster, uses nCluster to "better understand traffic on their site and optimize the experience of their social network members."

NCluster takes off-the-shelf, commodity hardware, such as Dell and HP servers, and turns them into an analytics database that can be easily managed and is capable of massive scalability, the vendor says. (Compare server products.)

MySpace, which wasn't available for comment, uses nCluster to analyze hundreds of terabytes of data generated by 110 million users a month, Aster says. MySpace wants to know with great specificity how users are consuming resources on its network, what videos are popular, and other useful information, says Aster CEO and co-founder Mayank Bawa.

"MySpace needed to analyze complete data sets -- not just samples or summaries," Aster writes in a case study. "Sampling would completely miss infrequently occurring but highly profitable patterns. . . . With Aster, MySpace is able to collect 100% of their Web traffic data for analysis without the need for sampling data."

Adding large amounts of storage to a data analytics system, often a complicated task, is automated by Aster software, Bawa says. Just add hardware, and click a button, in other words.

"The Aster system at MySpace is a 100-node cluster capable of analyzing 360 terabytes of data," according to the case study provided by Aster. "Additional nodes can be added quickly through one-click scaling as the traffic on MySpace.com continues to grow exponentially."

Aster was founded in July 2005 and has had customers in production for almost a year, Bawa says. nCluster pricing is based on the amount of customer data and starts at US$100,000. Bawa says Aster has a handful of customers besides MySpace, but isn't ready to announce any other names.