On Sunday night, Ellison, sans Etch A Sketch, essentially announced almost everything he once spat out with derision: The new Exadata appliance--X3--now runs magnificently large workloads in memory, a concept he called "wacky" last year in reference to SAP's Hana in-memory architecture. Not only has Oracle embraced the cloud, it offers every conceivable aspect of it, from infrastructure-as-a-service, to apps in a public cloud, to apps in a private cloud managed by Oracle--much of which, Ellison proclaimed, has been in the works since 2004. Oracle has put its flagship database in the cloud, virtualized it, and is able to run it multi-tenant style--a notion Ellison savaged to much fanfare last year at Oracle Open World in his backhanded slap at Salesforce.com.

Ellison explained away those last comments, saying he meant that Oracle doesn't believe in building in multi-tenancy at the application layer--that doing so tends to put data at risk, and breaks many crucial data management functions. By building multi-tenancy into the database, Oracle's cloud applications, which rely on the underlying Oracle database, are then, by definition multi-tenant. Problem solved. Companies like NetSuite and Salesforce.com, whose technology was built more than a decade ago, had no choice, he admitted.

The cloud, he said, was assumed when Oracle began its seven-year journey to rewrite all of its applications into what is now the Fusion suite. After all, even as far back as 1998, Ellison believed so much in the cloud that he and his partner, Evan Goldberg, started NetSuite. (InformationWeek interviewed NetSuite CEO Zach Nelson right before Oracle OpenWorld.)

There was no explanation of Ellison's commentary on the "wacky" idea of SAP's Hana, but he did manage to say that Oracle's new Exadata X3 platform had 4 TB of main memory, and 22 TB of flash memory, compared with SAP's meager half-terabyte system. SAP's Steve Lucas, executive VP of database and technology, noted in a phone interview with InformationWeek that SAP is already shipping a 100-TB, all main memory (that is, no flash) system with IBM. Lucas called Oracle's Exa-systems "Exaggerated, Expired, and Expensive," adding that there is little innovation in what Oracle has done, and that Oracle's customers will still have to cobble everything together on the Exadata system--everything that is already integrated and built into Hana. And that was the "calm" Lucas--he talked to InformationWeek one day after Ellison made his comment.

Oracle and SAP will surely duke it out around architectural and technology superiority, both in entertaining public commentary, and, one presumes, in front of customers, where it ultimately matters most.

We'll let that play out however it will (and offer blow-by-blow where we can), but I specifically pressed Oracle president Mark Hurd on the issue of technology innovation, in light of these seeming reversals, asking him to give Oracle a letter grade on enterprise software. "A," he said, without hesitation.

When asked how that could be the case when Ellison had so boldly dismissed nearly every concept it now embraced, Hurd noted that things like Oracle Database 12c, Fusion in the cloud, and Exadata's in-memory architecture don't get built overnight. In other words, all of these technologies have long been in the works.

Fair enough, but then why the earlier, public critiques? Why warn of the dangers of multi-tenancy in such a public way? Why didn't Ellison say last year that multi-tenancy at the application level was the real danger, if that's what he meant?

(To watch our full conversation with Hurd, tune into Valley View on October 24 at 11 a.m. PT.)

But it's probably unfair to ask Hurd to defend what his boss says in public, nor is it all that meaningful to scrutinize the past ... except perhaps to predict the future. EMC's Tucci talked about the next big trend: the software-defined data center. Not long from now, Ellison's a good bet to laugh off that trend, call it utter lunacy, and then kick off the 2013 Oracle Open World with a software-defined data center announcement.

While competitors and industry observers get lathered up, it's easy to lose sight of the outcome: Oracle suddenly has a strategy that transcends its database, never mind how it got here. Yes, Oracle's going to be just fine. The company continues to grow, stockpile cash, and, based on Oracle Open World attendees InformationWeek talked to, win over customers with its one-stop-shop approach. For example, Dan Drawbaugh, CIO of University of Pittsburgh Medical Center, and one of the country's most respected technology executives, is on an aggressive path to spend more than $100 million on a transformational healthcare analytics initiative with Oracle applications and Exadata as foundational elements.

Oh, we've heard plenty about customer struggles, especially on the integration front, especially about how green these new Fusion applications are. SAP's Lucas is convinced this is where the Oracle stack will fall down.

When I asked Oracle's Hurd whether his customers would also give the company an "A," as he did, on "excellence in enterprise software," he acknowledged that the company has its work cut out on the integration front. But he was steadfast in his belief that Oracle is on the right path.

And that path will hardly be a snoozer.