Is There A Place For a Premium JVM?

When Sun introruced HotSpot, they originally planned to charge for it. In the end they decided to give HotSpot away, but that was probably a difficult decision internally at Sun. And it was one of the best things Sun ever did for Java; the boom of Java & J2EE in 199-2000 might not have happened without a gratis, high-performance JVM. (At that time there were alternative high-performance JVMs, but all of them commercial or somehow encumbered - IBM JDK, TowerJ and others.)

It's fine to charge for extraordinary stuff like Mission Control, RT, or bare-metal VM that runs on hypervisor. Or premium support of course, like Sun did for a long time (JavaSE for Business). But charging for extra optimizations would be a disaster. People would immediately flock to competitors that offer their state-of-the-art compiler/VM runtime - from IBM (in WebSphere) to Microsoft (with .NET) - without extra charge. Even if they don't really need that extra ounce of performance that must be paid; it's a strategic issue, and people would immediately think "today Oracle charges for top optimizations; next year they will charge for common bugfixes; later on, even for security patches" and this kind of FUD might destroy Java.

I've dealt with Oracle for many, many years before they bought Sun. With the exception of their namesake database, which is excellent, their products are mediocre and overpriced.

I've worked with their sales staff, I've worked with former members of their sales staff and the products don't mean anything at all. They could be selling widgets or shampoo. They pad on half-baked features, provide lukewarm support and release buggy software as a matter of course. They're looking for iterative cycles to drive sales, keep customers on board and keep selling support contracts.

I hoped that somehow the open source underpinnings of Java would survive that attitude. But it's not. They've lost all of the heavy hitters from Sun, they've actively worked to disenfranchise the community and they're only just beginning.

They just removed InnoDB from the cheapest supported version of MySQL in an attempt to further monetize that market or drive customers upscale to more lucrative Oracle licenses.

Congratulations Oracle, you've managed to break through my complacency and seriously scare me about the future of Java. If it's going to end up as 20k/year support contracts, opening TARs for production failures and locking up docs behind pay walls, then so long and thanks for all the fish.

It's on you to prove me wrong. You've got about six months of leeway before OpenJDK needs to show what the future of Java will look like. I hope I'm pleasantly surprised.

However I am not sure if they can hide the optimization/bug fix/enhancement codes in the premium edition. The OpenJDK is GPL licensed and the premium edition is based on the OpenJDK so it must be GPL licensed too. Which means if someone buys the premium edition she can ask for the code and release it for free (like Red Hat Enterprise Linux is paid sotware but if someone recompiles the source, like CentOS it can be released for no cost).

Frankly I don't understand why people can't just see that the market has two legs: the "community" and "the rest of the world". The community is the smaller part. For what I can see, Oracle is going to keep a free, FLOSS OpenJDK for the "community" and a premium, paid version of "the rest of the world". While I don't have any experience of the "buggy" software Oracle sells and how professional are Oracle's salesmen, I see that Oracle sells a lot. So I assume It's not so strange that they could sell the premium edition without problems (probably, with its cost embedded in a global offering of a vertical stack).

In the end, they could be able to do what Sun always failed to do, that is directly monetize from Java. If Sun had been able to do, they would be still alive and kicking.

Oracle knew, that their lovely high priced monstrum runs 80% or something on Solaris, driven by a Java app. A huge amount of people had Sun servers. They wanted the stack, and they wanted it now. In conjunction with MySQL they got everything. The plan worked.

I think many people use too much anti-Oracle, anti-capitalistic 'reasoning'. But lets stick to the facts: who paid for JRockit in the past? There must be companies who did. I don't know even one. If Oracle joins JVM and JRockit, and leaves out the parts that make JRockit "so good they can charge for it", what does the JVM lose? Nothing.

It might gain in performance and quality in areas they can't monetize, since they don't need to handle two codebases. Currently everybody is hitting on Oracle, but besides raising the money for MySQL support (which was an expected move) and acting more like a classical US company (and less than a hang-out place), I can't see the big darkness everybody is picturing. 

Sun on the other hand, did a classical 'Microsoft': selling something, so they can crossfinance something else.Oracle wants you to buy their database, their servers. It would be a stupid move to put everything behind a $1000/CPU wall. The JVM (and the OS) drives the sales of the rest, which will easily bring in 10, 20x of that. But they don't want crossfinancing, they want the thing to be self sustainable. Simple basic business.

I don't like Oracle myself, but demonizing and painting black horizonts never provided any reasonable solutions. I doubt that 5% of those who currently cry foul will really start writing business software in PHP or C#. It will not happen, since most professionals don't deal with the business side of the environment they work in.