OpenSolaris Q&A

OpenSolaris.org Launches

Originally uploaded by sogrady.

Lots of things to talk about, but very little time today. Besides a full slate of regular responsibilities, I’ve finally got a new machine to configure. Shipped from Shanghai last week, my Thinkpad X40 arrived this afternoon so I’ve been busy loading the XP partition with necessary applications (Firefox, Spybot [1], etc), then shrinking the XP partition down to under 10 GB’s (it’s unreal how large it is by default), and beginning the long process of getting a new instance of Gentoo laid down.

Despite the schedule, however, I did want to be sure to take a minute (and actually, I have several as Gentoo’s at a stage where it’s installing the base applications, which will take a few hours at a minimum) to comment on the news out of Sun this morning. In case you missed it, the big deal is that Sun’s venerable and recently revitalized flagship Solaris operating system, after a long and arduous process, has been open sourced as OpenSolaris. Now before I continue, let me remind you as I have before that Sun is a client, so you’re welcome to read into the following comments what you will. On to the Q&A:

Q: What’s the best thing about the open sourcing of Solaris?
A: I may be in the minority here, but I think it’s good for open source more broadly. Conventional wisdom would have us believe that Sun is anti-open source (try telling that to the GNOME folks by the way) and that OpenSolaris is nothing more than an attack on the Linux operating system. Examining those charges, I find little evidence to support the first argument, but don’t really dispute the second. OpenSolaris is indeed intended to compete with Linux, and if you talk to Sun they’ll tell you so. They’ll happily run Linux on the new Opteron boxes that John Fowler’s group is pumping out, but they’ll lead with Solaris. How, you ask, is this good for open source?

Well, ask yourself whether having another viable open source Unix like operating system with a good pedigree and some shiny new features (DTrace, SMF, etc) behind it is good for a.) open source customers, b.) ISVs and c.) Linux? Obviously, I think it’s good, and for all three constituencies. It’s good for customers because they gain another viable choice, which will help if nothing else in negotiations with Red Hat and SuSE, and even Windows. It’s certainly good for Solaris focused ISVs because they now have an open source story to tell; for Linux or Windows ISVs, it may not matter, but may provide another choice. But what about Linux? Well, much in the way that Linux has become a personal trainer for Windows, OpenSolaris has the potential to push Linux to become more scalable, more reliable and more developer friendly (from what we hear, DTrace has really opened some eyes in the Linux community, and spurred development in that area).

I’m generally a believer in the notion that competition is good (except when it comes to standards), and for that reason alone I find the opening of Solaris interesting.

Q: What’s the worst thing about the open sourcing of Solaris?
A: The volume of questions and queries I now have to field that fall along the lines of will Linux kill OpenSolaris or vice versa? I’ve said it before, and I’ll say it again: such binary arguments are of little use, and are completely out of touch with reality. But don’t take my word for it, see what Linus himself had to say when asked about the differences between BSD and Linux:

I really don’t much like the comparisons. In many ways they aren’t even valid, since “better” always ends up depending on “for what?” and “according to what criteria?”.

Same deal here. Linux has strengths, Solaris has strengths. Why am I putting Gentoo rather than OpenSolaris on my new x40? Because it’s much better equipped for that task, and the community behind that type of implementation is far broader. What about the new two way, 64 bit server I’ll be colocating? That might very well run OpenSolaris, because that’s a role it’s very comfortable playing. Without question the two operating systems will be competing, but let’s try and stay away from the binary / one-will-kill-the-other arguments. As Linus says in the interview linked to above, “The world simply isn’t black-and-white, and I recognize a lot of grayness. I often find black-and-white people a bit stupid, truth be told.”

Q: What can we expect from the OpenSolaris community? How, for example, would it compare to the Linux community?
A: Well, taking the second question first, it won’t, not initially. Because of the emotions involved, people seem compelled to compare the OpenSolaris community to that supporting Linux on day one of the launch. I don’t see the point. My expectations are that OpenSolaris will have strong support in its traditional constituencies such as financial services and other Fortune 500 organizations, along with interest from pragmatic open source projects like Apache and Gentoo. Mainstream open source developers, however, are more likely to take a wait and see approach to gauge whether or not the new project is building any momentum. Given the technical strengths of the platform, I do believe that OpenSolaris will have staying power, but do I expect the OpenSolaris community to rival the Linux community in size in the short term? No.

Q: What’s the biggest challenge the OpenSolaris community is likely to face?
A: Well, probably the FUD that competitors will inject into discussions around OpenSolaris – but I think participation may be a challenge. On the call this afternoon, I asked how the OpenSolaris folks might deal with the “intimidation factor” with respect to contributions. Most open source participants have had a patch or contribution shredded at one point, but it’s a bit different with OpenSolaris given that a.) the code was just open sourced and b.) the brainpower of the people behind some of the pieces is immense (just have Bryan Cantrill or Michael Shapiro walk you through the new featureset sometime, and you’ll see what I mean). This could be an obstacle to participation for some would-be developers, but to their credit, folks like Bryan have done a good job of calling out and encouraging participation around projects like adapting DTrace to Java or Python. Then there’s the joint copyright requirement, which has been cited as an issue by a few people I’ve spoken with, but OpenOffice.org is one indication that that need not be a show stopper.

Q: What did you find most interesting about the launch itself?
A: The way that it targeted the most important constituency of all; developers. Back in April, I posted the following note:

Spoke to a vendor marketing representative yesterday who I won’t name (though they should feel free to identify themselves if they wish) that actually initiated a dialogue around del.icio.us, Flickr and tagging in general. First time that I can recall that’s happened. Very refreshing, and a good sign for the product line in question.

Well, that person was Claire Giordano, and the product line was OpenSolaris. Her hand is directly visible on the OpenSolaris.org page itself with links to developer-friendly services like del.icio.us and Flickr, and also in the lack of a big bang style launch, press release, etc. The focus has instead been on conversational and participatory launch mechanisms like the aforementioned services and an explosion of related blog entries. OpenSolaris, as with any other product, will ultimately sink or swim on its own merits. But if those merits will jointly determined by Sun and the community around OpenSolaris, it’s obvious that engaging developers using the means and mechanisms that they prefer is imperative. Given that context, I’d say that today’s launch was an excellent start.

Q: Any last thoughts on the license Sun chose, the CDDL?
A: That’s a whole other entry, and I have to get back to my Gentoo install, but in the meantime I found Linus’ take on licensing here quite relevant (link via Paul Murphy).

[1] On that note, does anyone else find it odd that Windows ships without a default interface for removing programs that run at startup? I constantly see friend’s machines that have 10 or even 20 applications running in the system tray, largely because there’s no centralized way – apart from third party apps like Spybot – to alter them. That impacts performance, particularly on older machines.