SAP, like its rival Oracle, is in the middle of a most interesting transformation. At the risk of implying a reptilian constitution, SAP is molting. It is still a software company, but it is trying also to provide a software suite in the cloud, for companies of all sizes.

As Oracle has acquired and then built software to compete with market leader SAP, SAP has been quietly (until now) building a new approach to data, long the province of Oracle. Hana is SAP's in-memory database, and its Oracle killer (if Oracle wasn't also molting), and it's de-emphasizing the database in favor of running the entire computing stack, from the hardware to the software, and everything in between.

We'll leave the debates aside for now. This is the story of SAP's journey. We'll be telling Oracle's story on October 24 on Valley View (11 a.m. Pacific time)--straight from Oracle's president Mark Hurd, and some of Oracle's customers.

InformationWeek invited SAP's co-CEO Jim Hagemann Snabe to join us for a fireside chat at our recent InformationWeek 500. I found him without pretense, with a tendency toward self-deprecation. My cynicism antenna is always at full alert, and if Snabe was acting, he's damn good at it. Let's not equate that with the conclusion that SAP's strategy is brilliant, it's just that it is rare to find an executive willing to give his company a grade of "C" on technology innovation (although he quickly assured the audience it was moving toward "B+" and eventually an "A"). He gave SAP a "B" on cloud and an "A" on big data.

He spelled out SAP's cloud roadmap, which is still a work in progress. I suggested that if the company's so-called cloud savior, Lars Dalgaard (the CEO of SuccessFactors, which SAP acquired last year), had called the SAP cloud baby "ugly," perhaps a score of less than "B" was more appropriate. He didn't fight that, but noted that Business ByDesign, which is SAP's cloud solution for smaller companies, had 1,100 customers. SAP just began shipping new cloud solutions for larger companies as well--it already had a head start with SuccessFactors in HR, and now in procurement with the acquisition of Ariba. But now Financials On Demand and Sales On Demand have shipped, he said, despite the absence of a formal announcement from SAP. "We want our customers to talk about it," he said, adding that they are "beautiful" applications--a point he stressed often.

And in this, Snabe was steadfast during our discussion, continuously referring either to customer deployments, or customer applications of SAP's technology. When I asked him where in-memory technology would be transformational, beyond just making queries faster, he said: "Simplification ... you'll solve problems you've never thought of before," and then gave an example around healthcare analytics, where a problem that would take two days and cost $100,000 per patient can now be done in one minute and cost $4,000 per patient. (You can hear the full answer starting at around the 24:39 mark in the video below.)

Snabe stated that by next year, the full SAP Business Suite will run on Hana, without any major changes; that is, it plans to run everything in memory, making it possible for customers to forego Oracle's database, at least in theory ... surely that won't happen overnight, if ever, but the ability to run transactions and perform analytics on a single architecture is promising enough to raise a few customer eyebrows.

Meanwhile, Snabe said that if SAP is to win in the cloud, it needs to be more like Facebook in its agility and simplicity; but unlike disruptive competitors like Workday and Salesforce.com (I'm naming them; he didn't), SAP will be profitable, he implied.

He also implied that Oracle stretched the truth with regard to SAP's cloud progress: "Maybe it was my parents who taught me never to lie," he said in response to Larry Ellison's statement that SAP would have "nuthin' in the cloud until 2020."

Snabe wasn't shy about going after Oracle, saying "I've never heard a customer say 'I want a stack,'" in reference to Oracle's technology approach. In response to Ellison calling Hana "wacky," Snabe said that Ellison is right, that 10,000 times performance improvement is wacky. But Oracle's approach to in-memory computing is all wrong, according to Snabe, who compared it to a hybrid car, with two of everything, all of which must be carefully managed ... a more complex approach than Hana.

I haven't given it all away here, trust me. Snabe and I covered quite a bit of ground, and then he faced a few tough questions from customers in the audience; he even managed a quick riposte on SAP's ongoing legal battle with Oracle. Watch the full interview in the video embedded below.