Technology speeds Linux database access

Some applications of high-performance computing require large databases and the ability to get information into and out of them quickly. As the Linux operating system matures, some developers are trying to tackle specific aspects of computing tasks.

At the LinuxWorld Conference and Expo earlier this year, officials from Objectivity Inc., a database development firm, announced that they have been able to load

32 terabytes of interrelated data objects per day into an Objectivity database using Irix, the version of Unix from Silicon Graphics Inc., and hosted on an SGI Origin server.

Objectivity officials are now trying to duplicate that feat on an SGI Altix server running Linux, said Leon Guzenda, Objectivity's chief technology officer. Company officials say they expect Altix to perform even better.

The capability, aided by SGI's NUMAflex shared memory architecture, matters because streaming large amounts of complex data with many relationships among the data objects is expensive and difficult, according to Objectivity officials.

"If you look at the vast amounts of data that have to be stored in a single instance, and that have to be looked at and interpreted in real time, these are areas [in government] that clearly would exploit this type of capability," said Tony Celeste, senior director of the defense and intelligence sector at SGI. "There are traditional programs in the defense and intelligence community as well that could exploit this capability."

The scientific market also needs the ability, he said.

In many government settings, "the needle in the haystack needs to be found across separate silos of information, and then fed to analysts in real time," said Jay Jarrell, president and chief executive officer of Objectivity.

Richard Winter of the Winter Group agreed that the government has a long-standing and pressing need for database technologies. "It's always been there," he said. "It's often pretty quiet, but as far back as 1975 I was working with the government on very large database applications. They've always been prominent, on the leading edge."