Illumos launched as OpenSolaris derivative (The H)

Java 1.6 will remain free. There's nothing they can do about that even if they want to, as it's dual-licensed under the GPL and CDDL. They can stop distributing it, but not stop others from distributing it.

There are some semi-core JARs that are under non-free licenses, like the JavaMail API and JPA API JARs. They're free enough that I don't think Oracle can block their distrubution, though, and I don't think anything would stop them being rewritten (quite easily) using the specification documents for those APIs.

<rant subject="java">

IMO, the biggest problem with Java isn't Oracle. It's Sun having gone totally off the rails with the language. Lambdas and true closures (for Java 7) are great, sure ... but in a language that lacks any kind of built-in properties and forces you to hand-write accessors ("getters and setters") they seem kind of the wrong thing to prioritize. One gets sick of hand-writing accessors and adding PropertyChangeSupport very, very quickly. Then there are the debacles of annotations and generics in Java 5, both of which would've been great if they weren't hobbled by (IMO over-strict) backward compatibility requirements. They were built by small, underfunded teams on rushed deadlines who weren't willing/able to accept external help, which caused them to be thrown out the door before they were really done, too.

I quite like the core Java language and the VM, but it's not going in a good direction, it's hobbled by backward compatibility requirements, and the fragmentation above the language level is horrifying. JSF/Wicket/Tapestry/Struts/.... web frameworks, TopLink Essentials/Hibernate/EclipseLink/JPA-API .... ORMs, etc. It's often really painful to figure out what's suitable for a project, and most of the tools half-solve a problem while creating more. The way most people seem to try to "fix" that is bolt another layer on top, another abstraction with its own problems, rather than fix the original framework or tool. "Enterprise" Java is a scary, ugly place that seems to exist to keep mediocre programmers in jobs.

I am a mediocre programmer myself, but I like to actually get productive code written rather than write the tools to build the data-transfer-objects to talk to the facade to access the data access layer to interface with the ORM to generate the queries to actually talk to the frickin' database. Argh!

</rant>

Others know very well they need Linux to achieve their profit objectives. (I'm sure IBM does a nice profit on Windows systems too).

Oracle is much too profit-oriented to wander around changing roadmaps every year and dropping product lines that earn hard money in the pursuit of phantomatic schemes like Solaris everywhere.

In case you haven't noticed, all the manufacturers that wished to renew their Solaris x86 agreement had to sign up on Oracle Unbreakable Linux too.

No I really don't think so. I know the story about how Oracle was suppose to crank out these super fast database monster machines since now they can optimize everything from the hardware on up. But everything Oracle has done so far points to them simply not caring about that approach.

I think they will continue to hack around with Solaris and support Sparc as long as people will be willing to keep paying the support contracts, but they are not going to put a lot of money and effort into that approach. Organizations will continue to want Solaris and Sparc because that is what their applications require, but I really doubt anybody is aiming to revitalize the Sparc platform and make big new deployments on it.

I think what they care about is their vertical application stacks. A OS and hardware for them is simply a mechanism to host Java VMs and provide a way to interface with storage in order to run database software. That's just about it.

Does Java/Solaris really provide that big advantage over Java/Linux to justify putting the time and effort into maintaining two enterprise operating systems?

> Solaris is crucial for that - for the same reason AIX is crucial for IBM. Not so with Linux, btw, and we can expect Oracle to start doing what IBM is doing, i.e. push its own system instead of Linux.

Well Solaris/Sparc is certainly no AIX/POWER. IBM POWER trounced Sun in the high-end and Solaris/Sparc is no longer competitive. Poring money into the hardware platform and trying to compete with IBM head on is one of the reasons Sun is dead now and I do not think that Oracle is aiming to repeat history.

It seems likely that Oracle's approach to scalability now seems to be the clustering approach, which is Linux's bread and butter and cheap x86-based hardware has a clear advantage. The advantage to custom hardware in this situation is to provide high-speed interconnect, improve the hot swapping of hardware, improve cooling, and increase density... small fast drives, blade servers, and the like.

I think your misreading IBM a bit also. IBM is extremely mercenary and I don't think that IBM really cares what OS or what hardware your using as long as you get it from them and your paying the support contracts. They will recommend you use AIX or Linux or Windows based on your requirements and your ability to pay. I am sure that some IBM goons push AIX, but that may have to do with their sales commission more then anything else.