As you may or may not have heard, Oracle has announced its JVM strategy, or at least part of it. Action is needed because the post-merger company has one of those problems that you sort of want to have, in that it now owns two highly regarded JVMs: JRockit, which came along with the BEA acquisition in 2008, and HotSpot, which had been Sun’s. Say what you will about Oracle, but they don’t necessarily fall prey to “not invented here” syndrome, and are generally open to using acquired technology if it betters something already in their portfolio. According to ex-Sunnie Mark Reinhold, Oracle’s principal engineer, the ultimate goal is a JVM that combines the best of both companies’ machines — with JRockit’s mission control features and HotSpot’s performance quality, for instance. (You can watch Reinhold’s webcast on the subject or see the coverage at The H, The Register, or Datamation.) One can’t really quarrel with the decision to take the best of both machines and make something even better, though as Reinhold notes — “Customers have things in production on both, taking advantage of specific features in both. We’re not going to cause an earthquake and make systems fall over” — people depend on both, and adopting one would have at least spared half of Oracle’s customers from a transition. Still, you want the best for the long term. And indeed, the long term is what we’re talking about here: Reinhold puts the target for the merged VM in the 18 to 24 month range, which is an eternity in IT time. (Was the thought of a Sun-Oracle team-up even a blip on the radar in January 2008?)One thing that I note is missing from this discussion is another virtual machine in Sun’s portfolio: the Da Vinci Machine, which is aimed at making dynamic JVM languages perform better. Obviously, it’s an entirely separate project — the JRockit/HotSpot combo will be aimed at mighty servers powering Java EE apps that need top-of-the-line performance, while the Da Vinci Machine is still a bit of language-nerd research. But in the long run — and in IT time, what is 18 to 24 months if not the long run? — those high-powered servers might well need improved capacity to run dynamic JVM languages. I wonder if any of the work in Project Da Vinci will find its way over to Reinhold’s group. Technology Industry