Software is a shape-shifting business. The product isn't tangible, so the customer has to be convinced of the constancy, the ongoing commitment of the provider. Then the product has to take on a life of its own, anticipate the market and add attributes in a timely fashion that match where people want to go. Sun is capable of understanding this for several months at a time. Then it gets distracted and goes and picks a fight with someone that it thinks has an unnecessarily large share of the industry's customers. I was reminded of this when I asked Jim Sinur, Gartner analyst, whether Sun has got it right this time around. Sun did a good job championing Java and establishing, under pressure, a vendor community process that shared responsibility for Java's ongoing development. Now it's adding integration software from SeeBeyond to a middleware stack. That's a new addition to open source, not available elsewhere, and that's the kind of open-source initiative that has a chance of succeeding.
"Sun will be able to draw in some more hardware and operating system sales because ICAN [the SeeBeyond acquisition's software] has been added to the stack. This will be a short-term boost. Sun does not have a great track record with software and will likely lose interest, like it did with Forte," Sinur said in an E-mail response.
Ah yes, Java tools, I had nearly forgotten. Sun's performance as a software company is good when the software is tied to its hardware, the way Solaris is. It's spotty when it's one or two steps removed from the hardware, the way Java development tools and Java middleware are. Sun charged in with Java tools early, then retreated; it charged in again through the acquisitions of NetBeans and Forte, then seemed to slink away. Is Sun really in the Java tools business? Of course, answer Sun officials, but that answer is less obvious to many observers.
In the first round Sun, to compete on tools, needed to shift resources from hardware, or at least from Solaris, to make a determined effort. Instead, in the mid-1990s, resources went into slanting the releases of Java development kit and the Java Virtual Machine in favor of Sparc/Solaris (as a hardware vendor might be expected to do).
That incurred the wrath of partners who were needed to help establish Java in the enterprise. Sun straightened out these problems, established a level playing field and went on to confront Microsoft over changes in Java on Windows. But with these distractions, it seemed to lose track of its own Java products.
By making a competent middleware stack available as open source, Sun has a strategy that leapfrogs the competition rather than chasing after it. I'm convinced that Schwartz is making the right move in part because Sun needs the constancy of an open-source community around what it produces. Such a community would be persuasive to customers that no matter how Sun's goals change, they would own the source code and have support for it.
Against the odds, Sun is succeeding with Solaris as open source. It's rallied Solaris developers outside Sun, generated a vigorous community, and slowed the incursion of Linux in the customer base. But no one can say for sure at this point whether Sun's combativeness and the natural, peacekeeping tendencies of open-source communities will prove to be a long-term match.