Oracle's CEO Larry Ellison didn't waste any time before making big announcements during his welcome keynote address at the Oracle OpenWorld conference in San Francisco in late September. Ellison kicked off the event with an introduction of several new hardware systems, highlighted by an in-memory database technology.
Ellison demonstrated Oracle's new capability, showing off the new product against Oracle's 12c database system. The new in-memory option promises 100 times faster query processing with no changes to applications and an increase of double the current transaction processing rates. With the in-memory solution, transactions run faster on row formats, while analytics do better on column formats, he explained. This "breakthrough dual format" for both row and column in-memory formats for the same data, he said, is the reason the solution can deliver results at such high speeds.
Ellison also explained that there is no need to put the entire database in-memory. Customers can choose to scale out or scale up with symmetrical multiprocessing boxes (software that enables a single virtual machine to use several processors simultaneously) without the added complexity of transferring the entire database—a feature that he said distinguishes Oracle's in-memory solution from SAP HANA.
SAP claims the company is still missing the mark. "They are still trying to make queries run faster, but missed the chance to simplify the data management at the same time. SAP HANA has been delivering real-time performance to our customers in real-world environments for years," a company spokesperson told CRM.
"The good news is that you are now a believer in in-memory databases," said SAP Chief Technology Officer Vishal Sikka, claiming a four-year head start. "We have already changed the game, and we are moving on to the next frontier."
Mark Hurd, president of Oracle, resented the comparison. "I don't think they're comparable. SAP HANA has to be programmed. What we told you about has nothing to do with all that," he said at a press conference during the event. "It's moving your current apps from here to there and flipping the switch. All the magic is below the database layer. This thing of where you have to go and write a whole bunch of new software is very hard and very complicated. So I refute the thesis it's even a comparison."
Analyst Ray Wang of Constellation Research says it's hard not to draw comparisons, however. "On some level, what Oracle is doing is very innovative, because they're giving their existing apps, database, and cloud customers an opportunity to transition to in-memory by flipping a switch, 'which is hard to do,' to quote Ellison."
Still, Wang all but criticizes Oracle's approach to innovation. "A lot of the time they see what new thing others are doing, think about it, and eventually just cave in and do it as well. They might make some innovative changes, but by that point, they've dropped all the costs against it."
Meanwhile, Ellison also unveiled Oracle's M6-32 "Big Memory Machine," touted to be the fastest in-memory machine in the world. It's available immediately, and holds double the cores of the M5 it replaces, hosting 32 terabytes of DRAM memory.
"This thing moves data very fast and processes data in-memory very fast," Ellison said, explaining that the M6 has double the system bandwidth of the biggest IBM system and costs less than a third of IBM's.
As for the future of Oracle's development, Wang says the company's biggest challenge is keeping up with internal growth. Through a series of acquisitions, Oracle has found itself in the midst of innovative companies under its umbrella. Now, Wang posits, the question becomes whether or not Oracle itself can keep up with the pace these innovators set.
"For Oracle to win the innovation battle," Wang says, "the company must win over the mindshare of the Oracle customers by acquisition."