Oracle executive chairman and CTO Larry Ellison's OpenWorld keynotes are often made memorable through healthy portions of competitive bombast delivered in his witheringly sarcastic style. Ellison did not disappoint fans of such rhetoric during his second OpenWorld 2016 keynote on Tuesday, but nonetheless made a strong case for Oracle's cloud advantages over Amazon Web Services.
Point number one: AWS isn't optimized for running Oracle's database, Ellison contended. "Why do I say that? It's easy to say that, what would you expect me to say?" Ellison joked, before flashing a slide describing a series of benchmarks Oracle ran against AWS using the Oracle database and AWS's native Redshift and Aurora databases.
Oracle's cloud runs analytic workloads 24 times faster than when Oracle's database is deployed on AWS, while transactional workloads run eight times faster, Ellison said. The benchmarking parameters and related downloads are available on Oracle's website. "Amazon is welcome to publish their own results. If they can beat our results we'll put that in," Ellison declared. "We don't think they can."
It's not just about pure speed, he added. "That has a huge cost implication," Ellison said. "The cost implication is that what we do in one hour at Oracle you do in 24 hours at Amazon. These aren't just performance differences. These are enormous cost differences."
Other shortcomings of Oracle on AWS include the lack of support for RAC (Real Application Clusters), Oracle's scale-out option. "You get Amazon's largest node, and you're done," Ellison said. That makes an Oracle database on AWS a single point of failure, he argued. Should there be an issue, "you are down and you are down hard."
Oracle's database blows away AWS Redshift as well, with more than 100 times better performance on basic analytics with half the CPUs, Ellison claimed. Redshift is based on technology Amazon acquired from ParAccel, which developed an MPP (massively parallel processing) database using a fork of PostgreSQL. That fork, Ellison noted, dates to 2005 and many mainstream Postgres features aren't supported in Redshift.
"Amazon is totally proprietary. It is not an open-source database," Ellison said. "The only people working on Redshift is the team at Amazon. The only place on earth Redshift runs is Amazon. If you build an application on Redshift uh, well, you're going to run it forever on Amazon. You are locked in, baby."
Redshift, in fact, lacks features that Oracle shipped for its database as long as 30 years ago, Ellison claimed. He went on to heap similar scorn on Aurora, AWS's MySQL-derived database, but still wasn't out of ammunition.
"We run everyplace. We give you lots of choices. The hardware of your choice or a specialized machine like Exadata," Ellison said. "AWS is more closed than an IBM mainframe."
Finally, Ellison recapped the details of Oracle's next-generation IaaS, which the company says delivers far more compute, I/O and storage than AWS's most powerful instances, for 20 percent less money.
Analysis: Ellison's Rhetoric Versus Market Reality
Oracle has a long road ahead of it catching up to AWS on a revenue basis. It logged $171 million in IaaS revenue for the quarter, an increase of 7 percent. In contrast, Amazon reported $2.89 billion in revenue for AWS in its most recent quarter, a rise of 58 percent year over year.
And while Ellison's trash talk was as entertaining, as usual he also left out some key matters when outlining the competitive stakes.
"The key difference is cost for large workloads," says Constellation Research VP and principal analyst Doug Henschen. "You'd have to dig into the details of these published benchmarks to figure out what the database license would cost compared to Redshift or Aurora. There were no cost comparisons for like capacity, but the stats suggest that you'll need less Oracle capacity to get like performance."
"As for 'low cost,' Exadata Express is a cheap new option, but it's not designed to support enterprise-class workloads," Henschen adds. "All of his points on 'only on AWS' are all true, but Ellison is not addressing those who are already using AWS here. I think the intent here is to discourage existing Oracle customers from even thinking about trying AWS."
The bottom line is that Oracle has to boost Oracle database and middleware adoption in the cloud, Henschen notes. "It was the key to Oracle successfully moving into all the application categories it's now in and it will also play that foundational role on in triggering app choices in the cloud. In other words, the stakes are high on challenging AWS."
24/7 Access to Constellation Insights
Subscribe today for unrestricted access to expert analyst views on breaking news.
