During the conference call with Wall Street analysts this week, covering its fiscal fourth quarter financial results, Mr Schwartz ran through a long list of statistics where Sun was growing market share, and then during a familiar rendition of the two horse race story, which posits that the server market has come down to two players, IBM and Sun, Mr Schwartz blurted out that Sun was considering porting its Solaris 10 operating system to IBM’s Power family of servers and workstations and Intel’s Itanium processor. Yes, the very same Itanium that Sun created Solaris 9 for and then spiked as its top brass called it the Itanic for years.
Sun has correctly identified that Solaris is what will drive its future, and maybe porting it to Power and Itanium makes sense. Since Solaris 9 was already running on the Merced Itaniums, moving Solaris 10 to the Itanium 2 architecture might be a relative snap. And rubbing sand in IBM’s eyes that it supports Solaris on more platforms than IBM does with its AIX variant of Unix would be something of a PR coup, as would being able to run Solaris on Hewlett-Packard’s Itanium servers. Getting IBM’s cooperation to run Solaris on its hypervisor layer on its iSeries and pSeries servers will be somewhat problematic, and similarly, HP is not going to be helpful in showing Sun how to load Solaris into the physical and virtual partitions of its Integrity server line. Neither will be easy, but both are probably doable.
However, the incremental sales that Sun might get in Solaris sales would probably be small, and the costs of maintaining the versions for Itanium and Power could be a lot higher than the money they bring in. It could be a great deal of fun, nonetheless, and as Sun has done with Java, it could make a lot of noise that only turns into money a lot further down the road. Being able to say that Solaris will run on any mainstream server in current production is worth something. The question that Sun must be struggling with right now is, how much?
The real issue seems to be how much Solaris is running side-by-side at IBM and HP shops, and how much of it is slated to move to AIX, HP-UX, Windows, or Linux in the coming years and how much is expected to stay put on Solaris. Customers have been fond of Sun’s Sparc servers, even though they have been perceived as pricey in the past few years, but they almost universally love Solaris with something akin to zealotry. Maybe Sun can sell a lot more Solaris on other people’s iron than seems possible at first glance? Maybe Mr Schwartz was throwing out ideas just to demonstrate, once again, that he is thinking outside of the box?
Or maybe we should just get used to these kinds of statements from Sun’s president. At the Sun Network conference in Shanghai in early June, Mr Schwartz said in a press conference with John Loiacono, head of Sun Software, that the company was definitely taking Solaris open source. The press relations people at Sun were a bit surprised to hear this, particularly since open source Solaris was in the early development phase.
What Sun will probably do to make Solaris open source is mimic the Java Community Process, which allows outside contributions by propellerheads and bittwiddlers, but which also keeps control firmly in the hands of Sun. It is not entirely clear how open Sun can make Solaris. Sun is based roughly on the BSD tree of Unix and Sun has also got all of its licensing in order with the SCO Group concerning Unix intellectual property, but making Solaris open source exactly like Linux seems to be impossible because of the intellectual property issues. Sun will have to control how Solaris is copied and distributed even if it does allow outsiders to contribute to Solaris development. It may even turn out that the open source Solaris community gets the job of porting Solaris to Itanium and Power and supporting it as well.
