Skip to main content

Java as an incentive for buying servers and operating systems

Posted by joconner on March 9, 2009 at 1:19 AM PDT

Hearing all the frustration surrounding the unavailable JavaFX environment on Linux and Solaris, I remembered a time -- maybe a decade ago -- when Sun actually hoped it could produce the fastest, best Java for Solaris OS and Sparc hardware. The idea was that customers would actually purchase Solaris and Sparc if Java performed the best on those platforms. You can bet that engineers made sure that every Java feature worked perfectly on Solaris/Sparc back then.

It made sense really. Sun makes Sparc hardware, Sun makes Solaris, Sun makes the very popular Java platform. Sun had no trouble convincing the world that the world needed Java. Seems reasonable to think that customers would put up some cash to put a strong back-end together for their language and JVM. If Java runs best on Solaris/Sparc, then maybe they'd buy Solaris/Sparc. Both Solaris and Sparc hardware had a decent price tag too, something substantial enough to pay the bills and engineers.

Would the same strategy work now? If you just knew that Java would perform amazingly better on a Sparc with Solaris, would that convince you to include Sparc and Solaris in your technology stack? Maybe? As an engineer/developer, if you knew that Solaris ran perfectly on your Dell laptop, would you run Solaris there too? If you knew that the really fastest, best codecs for JavaFX were available on Solaris and not Mac OS X, would you give Solaris a second look? Maybe?

One, maybe two more questions. Do you get a sense that Java actually runs best on every other platform? Are Linux and definitely Solaris second class citizens compared to Windows and Mac OS X? These days, instead of making sure that Solaris hosts the best Java performance, does Sun target other platforms first? And if so, how does that make Sun better? How does a better Java experience on Windows or Mac OS X make more money for Sun?

Related Topics >>


Come on: Java on the Mac is a disaster. Java 6? Only on Leopard. Java 7 pre-binaries? Nowhere to be seen. On linux on the other hand Java runs perfectly, GlassFish works excellent and NetBeans is a fantastic IDE. JavaFX might be different - but I do not use it. So for me Java on linux is just fine as it is. Right now I only use OpenSolaris on a client every now and then. So I cannot say much about it yet. But given that JavaFX is somewhat targeted at designers I do not expect Solaris (and Linux) to be of utmost priority here. And waiting for Apple is no option if Sun wants to push JavaFX as a RIA platform. So I think Sun is simply setting priorities according to the market requirements. That said I agree that simultaneous releases would be better.

I think Solaris has some very compelling features, but until recently it wasn't usable as a desktop OS. There are still some issues with 2008 11 which I've blogged about, and I hear most of them will be addressed in 2009 05 release (soon). The compelling features for me are: containers, zones, resource manager, rbac, SMF, ZFS, DTrace, and binary backwards compatibility for something like 7-10 years. I want to learn all of this while using OpenSolaris as my desktop so that I can later use it on servers instead of Linux.

The only differentiator I can really think of on that score is dtrace. If VisualVM and the NB profiler had really sharp tooling for dtrace, it might be enough to sway a few folks. I run ubuntu at work. If my userland experience were just as good in solaris (does it have as wide a package selection? How well does dual monitor work?), I might give it a serious try for dtrace.

The only chance for Solaris is to offer everything the others offer plus something else :) 1 missed feature is 1 missed system in a very competitive market..