Oracle Plays Catch-Up With New ‘In-Memory’ Database Technology

Oracle CEO Larry Ellison talked up the release of Oracle’s new in-memory database technology Sunday night in his opening keynote of Oracle OpenWorld in San Francisco, but was oddly non-confrontational when it came to trash-talking the biggest potential competitor in this space, SAP.

In-memory databases—such as SAP’s rival HANA database appliance—use memory instead of disks for data storage. That means that they perform very well in high-volume environments where response time is critical, since access times and database requests are typically much faster.

See also: SAP’s HANA Deployment Leapfrogs Oracle, IBM And Microsoft

The in-memory option for Oracle’s 12C database was one of the highlights of Ellison’s keynote. The CEO promised it would deliver a hundredfold improvement in analytic query speed and double throughput speeds for transaction processing. Ellison characterized this as “ungodly” performance.

Ellison claimed the feature can simultaneously handle transaction data as a row-store database while storing analytic queries as a column-store database. This dual-format setup supposedly offers speed advantages above and beyond the normal boost you would expect by keeping data in a system’s memory rather than accessing it from a disk drive.

To bolster Ellison’s point, a demo of the in-memory system queried seven billion rows of data in a single second, as opposed to five million rows per second in a disk-based database system.

Not only is in-memory an option for 12C databases, but it is also the heart of Oracle’s new Big Memory Machines, the SPARC M6-32 and Oracle Supercluster M6-32. According to a press statement from Oracle, “SPARC M6-32 servers with up to 32 terabytes of memory and up to 384 processor cores can run entire applications and databases in-memory to deliver unprecedented performance. Oracle SuperCluster M6-32… integrates SPARC M6-32 servers with Exadata Storage Servers optimized for Oracle Database performance.”

Oracle Plays In-Memory Catch-Up

Oracle needs to make a big splash in this space, since SAP’s HANA in-memory system has already garnered a lot of attention in the enterprise space of late, and IBM and Microsoft are working on their own in-memory solutions right now, too.

That may have been why Ellison was more subdued than usual. He didn’t call out SAP HANA explicitly, presumably because doing so would give attention to a database appliance that has been giving Oracle a run for its money. Some additional Exadata sales wouldn’t hurt, either, because it is not at all clear the product line is benefiting from all the product placement in the Marvel blockbusters.