% fortune -ae paul murphy

What's wrong at OpenSolaris

The OpenSolaris community just released the first complete "Project Indiana" edition of the OS - and in that process demonstrated why community enthusiasm for the product seems to have weakened recently.

It's a great product -everything from ZFS to a new package management system modeled on the blastwave process - and well worth examining carefully.

So why do I think there's a problem? As part of the release Sun offers to send anyone who requests it a free bootable CD - here's part of the headline bit for that page:

OpenSolaris 2008.05 for x86 is available to use free of charge and we can send you a CD at no extra cost. The delivery time may vary depending on your location, so you should consider downloading the CD image if you have a fast Internet connection.

The OpenSolaris 2008.05 Live CD makes it simple to boot to a fully functional desktop environment, including Firefox and Thunderbird, without the need to install onto your system. After familiarizing yourself with the OpenSolaris environment, you can then choose to install it onto your disk. Once installed, you can connect to the OpenSolaris Package Repository to install additional software at OpenOffice.org

Support for SPARC hardware is unavailable at this time.

Notice that? No SPARC support: on an OS product backed by Sun.

And beyond that, what does the curious but uncommitted user see here? an illogical "extra" in the first paragraph and a sloppy mistake in the second - the package repository isn't on the OpenOffice site, they're trying to cite an example of what's available.

Why did this happen? I think it happened for the same reason that companies like Red Hat, IBM, and Microsoft have been having a wonderful time spreading FUD about the CDDL and the inevitability of some future switch to GPL3: Ian Murdock.

By all reports he's a great guy with an outstanding record of achievement, but it seems to me that trying to turn OpenSolaris into Debian II stands the entire Solaris community ethos on its head, introduces a desktop/x86 hobbyist focus into what should be an enterprise grade effort, and indirectly ratifies the beliefs of those who don't see Sun as serious about OpenSolaris - thereby playing into the hands of those who want dismiss OpenSolaris as a Linux competitor that's bound to fail.

The primary argument I hear in favor of what he's doing is a simple one: "the numbers" say his supporters, "favor x86" - it's what the customer wants, it's where the demand is, and it's where Solaris needs to be. The second version of this that I hear fairly often is similar: "developers", these people say, "want this."

My problem is that I don't believe either version. On the contrary I would argue that computing jobs which can be done on x86, can be done using Linux, OpenBSD, or MacOS X - and what that means is that you can use Solaris for those same jobs, but making it possible confuses the market and actually doing it is a lot like hanging a 24" lawnmower under a V8: possible, but a recipe for injury, error, and bad press likely to mitigate strongly against Solaris adoption where it really counts: on SPARC and PPC.

The developer argument is partially a consequence of this: of course they say "Yes" when asked if they want easier, cheaper, more Linux like, access to Solaris; but when sales push meets support costs an x86 customer who doesn't forcefully demand the Solaris version will get the Linux one - meaning, on net, that lowering the Solaris entry point for x86 developers will attract vociferous developer support while simultaneously making them less likely to actually sell Solaris applications to their customers.

So, bottom line, what? Simple: I think making OpenSolaris seem more Linux like has been a mistake for Sun - a better strategy would have been to focus OpenSolaris development on SPARC and PPC/Cell, support a source compatible x86 version, and commit real resources to bringing Solaris capabilities like dTrace and ZFS to the Linux community. Basically, I see making Solaris more Linux-like as dragging Solaris down market when the same effort could have made a far more positive contribution toward dragging Linux up.


Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.