Normally such a presentation would close with reflections from a few beta customer and/or analysts before going into a Q&A (either during or soon after the main event). Instead the Webcast abruptly ended without so much as a goodbye. When that happened I asked public relations reps what happened and told them I had plenty of questions (and I'm sure plenty of other reporters did the same). "We don't know what happened," the PR rep replied, "but here's the press release." And that was that!It was a strange experience -- so much so that, for me, it undermined the credibility of the presentation! For one thing, instead of just delivering a long list of technical specs and superlatives, it would have been nice to hear just what sorts of organizations and IT departments are looking for one device that can handle both transaction processing loads AND data warehousing duties (mixing the two used to be a no no due to performance concerns). Can the hardware be tuned to serve both needs equally well? And are the two different constituent groups for OLTP and data warehousing likely to want to share and co-administer the same infrastructure?
I'm not saying that one ultra-powerful new machine couldn't blow away all expectations on both fronts and save a lot of dough in the process. But Oracle didn't provide any business or IT context on the need for a combined appliance -- just speeds and feeds.
As for all those stats and superlatives, Oracle offered zero independent validation or opportunity for scrutiny of the claims. That left me wondering if the "better/faster than anything else" assertions were tied to fine-print qualifiers and asterisks. For example, it's certainly true that IBM, Teradata and Netezza "just can't match this," as Ellison asserted at one point, but is that only because these competitors just don't offer a machine intended for both OLTP and data warehousing. It's easy to win a one-horse race, but that brings me back to my question about why organizations are naturally going to want this one horse.
Ellison made many claims about query performance, I/Os per second, loading speed and so on. They sounded quite impressive, but again, the lack of independent perspectives and opportunity for questions left me wondering if the fine print would note that Exadata Version 2 beats this particular stat or that particular benchmark when considering performance within the confines the full rack's massive 15 terabytes of flash memory? One question might have definitively dismissed all qualifiers, but who knows when we'll hear more (though the close slide of the Web event promised an Exadata Version 2 appliance will be up and running at next month's Oracle Open World event).
If Oracle wanted to foster open discussion, it sure picked a bad time for the product launch. For one thing, the company is set to announce financial results after the close of the stock market today, so yesterday wasn't a good time for Ellison or any other Oracle executive to be having discussions with analysts or the press. For another thing, Oracle's acquisition of Sun has yet to gain regulatory approval in Europe, where one commissioner has apparently raised a red flag about the prospect of Oracle gaining control over MySQL. Ellison and Fowler wisely said absolutely nothing about the pending acquisition; rather, the joint launch was presented as if it were a partnership of completely independent companies (which is actually the case at this point).
Meanwhile, Reuters is reporting that Oracle has abandoned the hardware partnership with HP that was the foundation of its first-generation Exadata appliance. I'd normally worry about orphaned customers, but I haven't seen any hard figures on just how many of those HP-based appliances were sold.
The bottom line is that as with Exadata V1, Oracle is throwing a lot of claims around about Exadata V2, but without availability of Oracle executives for questions, reporters and analysts can't do their jobs and give you real assessment of this product. It took Curt Monash nearly three months to gather enough insight to deliver this review of Exadata V1, and the conclusion was that much was still unknown about Exadata V1 three months after its announcement.
I'm guessing that by the time analysts and press get real insight on Exadata II, we'll see more flash-memory-heavy appliances and new products harnessing solid-state disks. For now I'm trying to sort through the many details, and I'll be on the lookout for insight from experts who have been able to pose questions to Oracle executives and sort through the claims. Exadata 2 might just blow us all away with its impressive performance, but I also won't be surprised if it just keeps us all befuddled for a good six months.Yesterday reporters were treated to an unusual product introduction from Oracle -- at least as the event was experienced via the Webcast. For about 30 minutes it was steaming along like a vintage Oracle product launch... and then it all abruptly ended without customer comments, analyst insight, a Q&A session or even a goodbye...