Sun to Sell Servers with Intel Xeon CPUs

When asked how long Sun had a team working with Intel under the covers to produce a Xeon-supportive Solaris, Schwartz revealed that his company isn't subdivided into teams, and admitted that as a unified operation, any work that his broader team does that concentrates mainly on Sun's own SPARC architecture can lead to problems.

"Frankly, the fact that they were only working on SPARC microprocessors under-leveraged the talent they had that could enable us to get into new markets," remarked Schwartz. "So we have a unified systems team at Sun that builds all the systems we build. In that regard, along with Solaris - which is obviously more than two decades in evolution - we've been working an awfully long time in the same space. The question was, when were we going to really commit to build common products? And I think that relationship has been going on for awhile, because we've seen one another in the marketplace so often."

After Schwartz had stated he wants Solaris "to absolutely scream on Xeon, to blow everyone else in the marketplace away," he was asked, if he wanted a screaming version of Solaris, how come Sun didn't opt to support Intel's Itanium architecture? His response did not come with the typical Jon Schwartz candor; instead, it was laced with a few mixed metaphors: "We try to go after as much market as we can, but these are loosely-coupled and highly-aligned businesses. So we want to see Solaris succeed on all platforms on which it ships."

Here, it was Otellini's turn to rescue Schwartz, but in so doing, he may have revealed Intel's sore spot: "The very positive part of this relationship is the ability to work together to get the Xeon-based systems...I think it would be the wrong thing to do to reopen the religious war on Itanium," Otellini stated thoughtfully. "Itanium is a separate product line; right now, Solaris does not support Itanium. If they decide to support it, we'd love it; if they don't, it's a business decision on their side."

While Schwartz characterized today's announcement as a "game-changing event," reporters attending today's press conference may have been watching a different game: perhaps replays of the AFC or NFC championships on their video iPods. At multiple points, there were no questions pending on the floor with minutes to go in the Webcast; and at one point, Jon Schwartz started asking reporters questions to see if reporters might be interested in what was happening in front of them.

But while this may not be a total rewriting of the rules of the server market, today's announcement does indeed change much of the scenery, even if reporters present hadn't spent enough time in data centers to understand its meaning. There is now an open-source Unix that may have a leg up on open-source Linux in its ability to produce binary-compatible versions that scale up to 8P and, eventually, the 16P servers of the near future.

Sun's Jon Schwartz tried to explain: "The single biggest determinant of the success of a high-scale x86 system will be the popularity of the operating system that runs on a single-socket x86 system. You cannot start by saying, 'I'd like to build a 16-way x86 system, and oh, I probably need an operating system.' No one starts their business on a 16-way system; they all start their businesses and their projects on one-way...This is the recipe that we know well. What's led to the success of Sun systems business is the fact that we've had complete binary compatibility up and down the product set. So the fact that we're going to be in this space with our own systems...with an operating system that eats threads for lunch and scales beautifully, should give us a little bit of a boost that maybe some of the other players haven't had."