If you want more evidence that Oracle wasn't thinking about cloud in 2004, you need only look to Larry Ellison's now infamous September 2008 description of cloud computing as "complete gibberish." This year at Open World, Ellison admitted he had been critical of cloud in the past. But he also suggested that he had always felt that multitenancy could be a good thing if executed in the database rather than in the application. So if Oracle believed in the cloud way back in 2004, why didn't it start working on doing multitenancy "the right way" earlier than in a database set for release in 2013?
Rave: A Kinder, Gentler Ellison. Larry didn't exactly pull all his punches, but he seemed more magnanimous and elder statesmanlike (with a few exceptions noted below) than during last year's Open World. He pointed to NetSuite (a company he partly owns) andSalesforce.com as cloud computing pioneers. There were also more mentions (even logos) for third-party vendors, including SAP, used within presentations by Ellison and others.
[ Want more on Open World? Read Oracle Touts Cloud Flourishes, Trashes Salesforce.com. ]
It's as if Oracle got the message that the "we-can-do-it-all" bravado was getting a little carried away. The sales push on engineered systems was also toned down several notches. If the performance of these appliances is all that it's cracked up to be by Oracle, then they should be selling themselves.
Rant: Same Old Bogus Claims. There were the usual barbs pointed at competitors and fantastic claims about Oracle product performance. Oracle President Mark Hurd, during his 12-minute, sales-pitchey Monday "keynote," talked about Exadata routinely delivering "100 times greater performance." Greater than what, he didn't say.
Real-world reports on Exadata aren't nearly as lofty. During Intel's Keynote presentation on Wednesday, for example, Deutsche Bank's Marcus Praetzas, director, financial data warehouse, said Exadata had helped the bank cut the time required for a credit risk calculation by half, dropping from 33 hours on Oracle database running on conventional hardware down to about 16 hours. What's more, that performance was delivered despite the fact that data volumes had doubled. That's a great improvement, but it's nothing like the "100 hours cut down to one hour" advances touted by Hurd.
Performance varies with the workload, of course, but I heard similar stats from Tim Gerber, a DBA at GPS company Garmin. After moving to Exadata from Oracle databases running on SMP servers, Garmin found that its Oracle E-Business Suite database ran 50% faster, its supply chain planning database ran 20% to 50% faster, and it's month-end and period-end close processes ran 60% to 70% faster. Once again, that's good, but we're talking percentages here, not multiples.
Another claim that stuck out was Ellison's assertion that Oracle is the only vendor that is delivering applications on premises, in private clouds, and on a public cloud. Sorry, but Microsoft articulated that strategy two years ago, and it has been doing it with Dynamics CRM and Office 365 for more than a year.
Finally, as for the claims that multitenancy implemented at the application level is insecure, where are the horror stories about NetSuite, RightNow, Saleforce.com, SuccessFactors, Taleo, or Workday customer data being hacked? And since when are databases so bullet proof? Security publication Dark Reading has reported on plenty of Oracle database vulnerabilities, including a September report on Oracle database passwords being easily hacked. Search "Oracle database hacks" at Dark Reading and you'll find other examples.
So there you have it; Oracle is embracing the new, delivering important new products, and, in some ways, softening its engineered systems sales pitches and slash-and-burn criticisms of rivals. In other ways, it's the same old Oracle.