Oracle Open World 2013: Oracle launches in-memory database search

Oracle has launched an in-memory version of its database which promises 100-times faster query processing.

Oracle has launched an in-memory version of its database promising 100-times faster queries.

CEO Larry Ellison kicked off the Oracle OpenWorld user conference in San Francisco by announcing the in-memory option for the latest 12C database.

Ellison said customers would witness “ungodly speeds” which would allow them to access “real-time analytics and instantaneous results” from their databases.

He said the in-memory system combines the two forms for storing data in databases – in columns, or in rows.

“We’re maintaining rows and columns with transactional integrity,” Ellison said. “And the data is consistent between the two formats.”

Ellison said that, while the in-memory format option will improve database search queries a hundred-fold,  the rate of transaction processing will double.

The company has removed database indexing to speed up processing, as processing queries in-memory reduces the need for indexing data.

Ellison said every application that runs on a 12C database can exploit the in-memory option and the transition would be seamless.

“All of this technology is cloud ready and ideal for the cloud,” he added.

No comparison with Hana

In a press conference today with Oracle president Mark Hurd, Ellison stressed the database format option was not intended as a strategic offering to compete with SAP’s Hana in-memory database technology.

“I don’t like it when Oracle Exadata or our in-memory options gets compared with Hana, because I don’t think it’s comparable,” Ellison said. “Hana has to be programmed.”

Hurd claimed that, in SAP Hana’s in-memory product, writing new software is difficult and gets very complicated. However, using the in-memory database from Oracle requires no changes to the application, according to Oracle.

“If you’re an Oracle customer with a database, the option is to simply flip a switch – it’s a huge advantage,” Hurd said.

Ellison said: “Every application you wrote, every app you bought – it runs without change, the only changes is that it takes less time to run.”

Hardware: M6-32 Big Memory Machine

Oracle followed this announcement by launching a new server to complement the debut of its in-memory database.

Ellison claimed the M6-32 Big Memory Machine has twice the cores of its predecessor, the M5-32. The SPARC M6-32 servers have up to 32 terabytes of memory and up to 384 processor cores. The company claims these servers deliver applications and databases in-memory with exemplary performance. The SuperCluster version of the M6 will provide users with the basic benefits of Exadata software integrated compression as part of an engineered system.

Ellison said it would cost less than a third of rival IBM’s similar offering, while offering twice the memory capacity of IBM’s largest power server.

Backup and recovery

The last announcement of the evening was the Oracle Database Backup Logging Recovery Appliance. Ellison said typical backup appliances were not designed for databases.

Ellison said these systems treat databases as a collection of files and, in the event of a recovery, only restore them to the time your last backup was taken, losing data from any intermediate phase. He also said they take a lot of processing time to make copies.

Oracle’s system is designed to conduct a backup log, followed by logs of changes to the database, which require little processing power. Ellison said the system does not need to be in a datacentre with a high-speed connection and it can be backed up into the Oracle public cloud.

The system provides customers with a choice: “You can put it in your datacentre, in the cloud or a combination of both – you decide,” said Ellison.

Read more on Server hardware