OpenSolaris, Sun's open-source version of its Solaris operating system, gets its official kickoff today at Sun's CommunityOne conference in San Francisco.  And it's not Sun's attempt to knock Linux out of the box -- it's something a little subtler than that.

OpSol (the abbreviation I'll use since I can't well call it "OS") has been packaged to make it friendly to end users: the default ISO boots right into a live environment with the Gnome 2.20 desktop, and it comes packaged with a slew of other familiar-'n'-friendly FOSS applications.  But the desktop stuff is actually not the big draw, since that can be ported most anywhere.

When I talk to Sun fans about what would be most attractive about using OpSol right now, they all give me the same answer: ZFS, which is about as industrial-strength a file-storage solution as you're likely to get anywhere, open source or not.  What's more, instead of using a version ported to another platform, OpSol lets you use ZFS in its native kernel environment.  (One truly exciting possibility here: creating ZFS-based storage appliances.)

The big downside to OpSol: its licensing, which uses the GPL-incompatible CDDL.  It isn't impossible to mix and match CDDL and GPL software, but it takes some legerdemain -- either by compiling the GPL driver from source on OpSol, or by running binaries of all GPL code in userspace, separate from the kernel.  This latter option means a bit of a performance hit for certain things that would normally run in kernelspace.  That and the mere fact of this incompatibility might mean developers who've habitually used the GPL would be uncomfortable working with a different licensing scheme.

Again, I suspect the goal with OpSol is not to unseat Linux, but to provide a complementary development platform for many of the same things that can be built there.  The one analogy which comes to mind, much as I hate to make it, is how Microsoft is trying to push Windows Server 2008 as a great place to run open source software -- although WS08 and OpSol are poles apart, natch.

Sun's current strategy for monetizing OpSol is about what you'd expect: selling services and support.  The cost of the support isn't cheap -- it starts at several hundred dollars and goes up to a few thousand per system per year -- which sounds like a reflection of the kinds of professional customers Sun has in mind for OpSol.  Desktop and "just plain folks" users can always snag one of the derivative community-oriented distributions of OpSol that have already sprung up -- like Nexenta, a very nice bundling of GNU GPL software with an OpSol foundation.  I've snapped up .ISOs of Nexenta and OpenSolaris and plan on giving them a closer look.

The big question in my mind is whether Sun can natively cultivate the kind of community-development cultures that have surrounded other open source projects, both OSes and otherwise.  If they can't develop it natively, then there's a good chance it'll spring up around one of the derivative OpSol distros -- but I'd rather see Sun approach its prospective communities directly, instead of leaving that vital job to others.