“Is the S390 platform still a project under development at OpenSolaris?” asks a frustrated subscriber on the OpenSolaris on z listserve. She continues: “I’ve been trying to subscribe to the S390 list at LISTSERV@vm.marist.edu without success for a week, the website wasn’t updated in months, and documentation is sparse at best. Even trying to find hardware is impossible. WTF is going on?”
Good questions; I’ve been asking similar questions too. The Solaris on System z effort was conceived in 2005 and an open source project initiated in 2006. A project update was presented at the Gartner data center conference in 2007.
Nearly two years later, March 2009, David Boyes, Sine Nomine Associate’s chief developer, again presented a project update to analysts. He reported that build 100 was available for download and apparently was working. Boyes also reported that only one customer application, so far, required source changes but that some applications actually performed 45-55% faster on the System z (z9 EC) than SPARC. Sounded like progress to me.
And then nothing. The listserve went nearly dormant. Late this spring and early this summer gripes began cropping up on the listserve and in a LinkedIn discussion, following a question I posted about the status of the effort.
In a phone conversation with Boyes, I asked if anyone was actually using the code. Although he had listed a handful of large companies that had downloaded a version in an earlier briefing, he avoided naming any actual current users recently. Besides, they won’t talk publicly, he added. Hmm, this was beginning to sound more and more like vaporware.
In Boyes’ defense, this was a complex undertaking from the start, involving multiple parties: IBM, Sun, Sine Nomine Associates (SNA), and the OpenSolaris on z community. And the acquisition of Sun by Oracle brought to a halt whatever progress was being made.
Finally, SNA people responded to the LinkedIn discussion, writing: We’ve been spending our time working on the code and dealing with the politics of involving the largest and (until recently) the 3rd largest computer companies on the planet in a project that neither particularly has a warm fuzzy feeling about. It messes up their product plans and carefully plotted market measures, and they don’t like that.
As for available hardware to run OpenSolaris on z, SNA reports: There’s plenty of System Z hardware out there. There may not be hobbyist hardware, but there’s certainly hardware out there. What you need is hardware that is capable of running z/VM and has a z/VM license. The second is the harder one.
He continues: I’m working on getting a machine that can be made available to the outside world, but that’s not free, and the other hardware available is maxed out supporting our own bread-and-butter apps. Sun has a rather nice z9, but they don’t have a good way to make it available for such work.
So that’s where OpenSolaris on z stands. Don’t expect to be running those potent SPARC Solaris applications on a z anytime soon. And if it finally ever is ready for prime time, it may not be what you expected—classic Sun Solaris on System z. The effort is based on the OpenSolaris code base, not the standard SPARC-based Solaris (Solaris 10).