After months of silence, OpenSolaris supporters have had enough and launched the Illumos project. Described as a "spork" of OpenSolaris, rather than a true fork, Illumos is a misguided attempt to keep the Solaris legacy OS alive for another generation. Too bad it's doomed from the start.
Software freedom means no project with a community will go away, so long as they have the means and will to sustain it. That’s usually considered a good thing, but sometimes it’s less of a feature and more of a bug. Consider, for example, the sad case of OpenSolaris.
Since Oracle gobbled up Sun, it’s remained mute on the fate of OpenSolaris. Attempts to get someone at Oracle to comment have been fruitless. I’ve spoken to Oracle PR and some of the employees on the community side about OpenSolaris and the responses have been both off the record and totally discouraging. To put it bluntly, Oracle seems to have put OpenSolaris out to pasture and won’t even do its community the courtesy of making it official.
So a hardy band of OpenSolaris enthusiasts led by Nexenta have taken up the banner and are trying to save OpenSolaris with the Illumos project.
It’s easy to blame Oracle for the sad state of OpenSolaris, and it certainly owns a share of the blame. But it was Sun that held the reins too tightly and built a community process that was too wholly dependent on a single entity. The state of limbo that OpenSolaris sat in during the Sun acquisition and Oracle’s subsequent silence afterwards was entirely predictable. OpenSolaris was entirely too dependent on Sun for development and resources.
Now it’s being reborn, and I’m not sure that’s such a great thing. OpenSolaris has always been a bad idea. Not to detract from the quality of Solaris — the operating system certainly has its advantages and surpasses my favorite OS (Linux, in case that wasn’t obvious) in some ways. But software freedom and community were bolted onto Solaris at the last minute as Sun was gasping for air and trying to remain relevant in the face of Linux. Sun folks, well ex-Sun folks these days, like to talk about how good Sun was to open source — but conveniently neglect the years that Sun fought open source and Linux kicking and screaming until it was too popular to ignore. OpenSolaris was not a friendly attempt to embrace community, it was a desperate attempt to remain relevant while Linux ate away at Solaris market share. OpenSolaris was largely an attempt by Sun to steer developers and companies away from Linux onto its turf.
Sun eventually did many things right with regards to open source, but the company was a poor steward of projects that it held. Sun retained far too much control and was far too fond of reinventing the wheel. Consider, for example, the decision to create yet another packaging system for OpenSolaris rather than just using Debian’s package system even though it had already been ported and was usable. Whatever technical advantages the new packaging system may hold, it’s hard to imagine that they would have outweighed the benefit of collaborating and improving an existing system.
Linux Won
That, writ large, is why it’s unfortunate that there’s an effort to revive OpenSolaris. Everyone would be better off if the Illumos/OpenSolaris folks would get on board and help make Linux and its ecosystem better. Parallel development of Illumos/OpenSolaris wouldn’t be too bad, excepting the nasty licensing barrier that Sun erected to ensure that interesting Solaris technology couldn’t migrate to Linux. As it stands, that sad state will continue. Worse, OpenSolaris still isn’t fully open source — so the Illumos folks are going to have to spend quite a few development hours just replacing bits that are already existent in Linux like internationalization functions of Solaris LibC and a lot of device drivers.
Granted, switching teams to Linux would be a big pill to swallow for folks with a long history with the Solaris family and an investment in OpenSolaris. But trying to bootstrap a successful community on top of OpenSolaris, which was not a particularly thriving community to begin with, seems like a Quixotian effort. No doubt it will continue to plod along, but the odds of OpenSolaris ever attaining critical mass are poor indeed.
Spare me the arguments that Solaris or OpenSolaris have this or that feature, or do this or that better than Linux. When geeks learn that technical superiority doesn’t count for squat when it comes to market adoption or long-term viability it’ll be a happy day indeed. OpenSolaris is Betamax to Linux’s VHS. Linux is superior where it counts: adoption, ecosystem, community, and diversity.
It seems to me that it’d be a better world for software freedom and free *nix in general if the Solaris die-hards sucked it up and helped work on Linux. Failing that, FreeBSD. Either way, it’d be an improvement over trying to “spork” OpenSolaris when the effort would be more effective elsewhere.
Ultimately, software freedom means that OpenSolaris fans can do whatever they want with the open sourced bits of Solaris/OpenSolaris. One could argue that an open project is successful so long as the users and community are happy, and there’s a bit of truth to that. But really? It seems futile to me. The odds of Illumos gaining any sort of momentum outside the existing community seem minuscule.
As a hobby effort, there’s no harm in keeping an OS alive as long as is humanly possible. But supporters of software freedom are always going to be hamstrung as long as there’s no willingness to throw in the towel on failing projects to join in on more successful ones.
Solaris had its day in the sun, so to speak. It will certainly live on in server rooms for another decade at least while Oracle wrings every bit of cash out of it that it can. But Illumos just seems to be prolonging the inevitable.
Joe 'Zonker' Brockmeier is a freelance writer and editor with more than 10 years covering IT. Formerly the openSUSE Community Manager for Novell, Brockmeier has written for Linux Magazine, Sys Admin, Linux Pro Magazine, IBM developerWorks, Linux.com, CIO.com, Linux Weekly News, ZDNet, and many other publications. You can reach Zonker at
jzb@zonker.net and follow him on
Twitter.
Comments on "Should OpenSolaris Die?"
What a great opportunity to make Red Hat and Fedora even better Linux systems. All the pieces that OpenSolaris does better, can you see the improvements to Linux? And won´t the Red Hat and Fedora communities benefit from OpenSolaris experts? Give the OpenSolaris team the credit they deserve, and praise their contributions to the future of Linux.
OpenSolaris deserves a place in the OSS community just as much as Linux, FreeBSD, NetBSD, OpenBSD, FreeDOS and the other OSen.
Linux is not the only answer. Your article even points that out with FreeBSD.
I agree with the package management system. There should be just one. Make everything debian based and kill off RPM. Imagine all the time that would be saved if developers only had to automatically generate .deb. RPM probably doesn’t do anything better then .deb, right?
I’ve been using Linux since ’92. It’s my desktop at home & work. I also use Macintosh and Solaris at home and work. Heck, my home laptop is Ubuntu and my work desktop is Fedora. Diversity is good and each of those systems offers something the others do not have.
I wish Linux had ZFS & I’ll be happy if BTRFS is only 90% as good. I wish I could create Flash Archives (think Ghost while the system is running) or Live Update.
I wish Solaris had a patching and packaging system as good as Yum or Apt or YaST.
I wish Linux had OpenBSD’s pf network filtering. And that I could build a contemporary Linux system inside a VM with 32 MB of RAM with the full repository available.
But they don’t right now. Until then, I want all these communities to succeed. Choice is good.
“I agree with the package management system. There should be just one. Make everything debian based and kill off RPM. Imagine all the time that would be saved if developers only had to automatically generate .deb. RPM probably doesn’t do anything better then .deb, right?”
That’s a naive perspective to the packaging problem. Package maintainers would still need to recompile and re-package for each different distro to make sure the binaries were properly linked against the libraries available on the system and everything found it’s way into the correct paths following that distro’s chosen conventions.
The idea that you can generate one package and install it anywhere is something of a myth that was propagated by proponents of RPM. It does sorta work with some packages, but only if you tread carefully with dependencies (and even then it can make things ugly in your system).
Opensolaris is an industrial grade OS that has something to offer the community. Oracle need a good kicking for not being open and honest. However Linux (and ????BSD) also have flaws.
Sadly, the current forking of significant open source systems can’t be good long term. Larry Ellison’s famous mantra “all the wood behind one arrowhead” had sense and still has something to say today.
Diversity and competition is how nature produces better creatures and the same applies to human endeavour. I no more want to be dependant on Linus’ view of how things should be done than Steve Ballmer’s (sorry Linus for comparing good and evil ;).
When I said only one packaging system should be used, I was being sarcastic…..
“All the wood behind one arrowhead” was Scott McNeely, not Ellison
“When geeks learn that technical superiority doesn’t count for squat when it comes to market adoption or long-term viability it’ll be a happy day indeed.”
Really? This is pure dumb! You should have said:
“When non-geeks learn that technical superiority is what should count for squat when it comes to market adoption or long-term viability it’ll be a happy day indeed.”
Dude, other than that you have said tons of BS! Sorry, but diversity is what counts, Linux folks won’t do what they do not want to do. What if I wish a microkernel and Linux is the only free OS out there?
http://www.kernel.org/pub/linux/docs/lkml/#s15-4
They won’t do it! They even point to some possible alternatives. Those folks understand what you don’t: Linux is _not_ the only way to go.
STFU!
This statement is outstandingly succinct and eloquent:
“When non-geeks learn that technical superiority is what should count for squat when it comes to market adoption or long-term viability it’ll be a happy day indeed.”
Linux is an example of how the statement “good enough is the enemy of great” can be so sadly true. Don’t get me wrong; I’m glad we have Linux in place of, say, Windows server, but anyone familiar with UNIX history knows that it was a perfect storm of unfortunate legal troubles on the BSD front that allowed it to gain its foothold. NOT its technical superiority to other Unix implementations
Illumos is a VASTLY superior operating system in many very measurable ways, and to suggest that the people who are passionate about it should somehow just fold up their tent and march dutifully over to the Linux camp is insulting. These folks have put a lot of themselves into Solaris (and now Illumos) and I, for one, thank them.
Remember, most people are not paid for their open source contributions. You can’t just ask leave a project they believe in so they can go donate their time to the one you happen to like.
Isn’t it funny how OpenSolaris is now the underdog, to Linux’s more main stream success story. But couldn’t many of the same arguments have been made about Linux, not that long ago. I’m disappointed whenever I read articles like this, that make a case to discourage enthusiasts from working on something they really love. We should all encourage anyone who wants to put something of themselves into a community project, regardless of the imperfect nature of said project. So I say, “Go for it!” Nexenta Systems, and best of luck to Illumos project. There’s plenty of room in the pool and the water fine.
It’s a shame for ZFS, and Sun made a huge mistake when they erected the license barrier preventing Linux from adopting ZFS. Oracle will probably let ZFS die like they’ll let Solaris die. Such a shame!
I desperately need ZFS on many of our systems (and there is no real alternative, regardless of what people say), but I could do without OpenSolaris.
For the forseeable future I would still rather have Solaris back-ends in my server-room for NIS and NFS servers. They are much more reliable and easier to configure. I still have Solaris servers which run for more than a year without rebooting. The same cannot be said for Linux servers unfortunately. So the rock-solidness of Solaris is still a requirement at least for me. Linux is fine for desktop clients, but you can’t beat the stability of a Solaris box on a high-demand machine.
> The fuurte of (G)obbyget that man on planet GNOME (he’s been helping out a lot on gtkmm as well)
t6gl9x wlcmluuddhnv
Unlike Beta vs. VHS (or OS/2 vs. Windows), this is not an XOR situation, as several commenters before me have attested. This is more like a GIMP vs. ImageMagick kind of thing (I install both, and use each as needed on my digital photographs). Depending on the license under which Illumos operates (hopefully the same GPL2 license underpinning Linux), the DNA of each system will transfer readily. It’s one of the strongest aspects of compatibly licensed open source — the parts can all be reused as needed.
In fact, in our new virtual machine world, we can easily have “all of the above”.
As a former Sun employee and currently a sysadmin in a very heterogeneous UNIX environment where Solaris used to play a dominant role until recently, I must say that it saddens me to see what Oracle is doing with what is left of Sun. On the other hand Sun can only blame themselves for what has become of the company and I witnessed a lot of disputable decisions during my time there but have always considered Solaris to be a very mature and reliable product. OpenSolaris is likely to go down the drain but mainstream Solaris is not that far away from sharing a similar fate. Solaris estate in my company has sadly diminished very considerably these days in anticipation of the inevitable in favour of Linux and AIX. We’ve all seen this story with other great operating systems that are now past but I never thought I would witness Solaris share the same fate. I don’t want to spread any gossip here but from what I hear from my former colleagues future isn’t bright for Solaris and that’s too bad. As many of you have already emphasized – diversity is what drives innovation and competition and something users can only benefit from. The day when we are left with only a handful of OSes will be a very sad day indeed.
Contrary to what cweberusa said, I believe ZFS will live on – after all, it under-pins the “Amber Road” Unified Storage (as does OpenSolaris).
I wouldn’t be at all surprised if Oracle were to start promoting a “database server out of the box” – in other words, an O/S + DBMS + storage solution (appliance?) ready to use from first power up.
Here’s what people in the open source community fail to recognize. Sun and now Oracle have no interest in making all of the IP inside Solaris free (for copying) to the world at large. Sun has given the world a number of “free” things from which we all benefit (NFS) but never been particularly good at making money from them (NetApp makes more from NFS that Sun ever did). At the opposite end of the spectrum is IBM, or MS, where everything is held back, protected as IP, and yes, some of it is able to make money because it isn’t used by everyone (at least not legally). Solaris is not GNU licensed for a specific reason, to protect ownership of the IP. If it was GPL’d, Sun/Oracle could no longer control it and would basically be functioning the same way Redhat is for Linux. There is no requirement on a company to open source any of it’s IP. However, consider that while not great for sharing of complete source, it is akin to a closed source operating system in which you have total visibility into how things work. Imagine MS Windows where you can freely browse the source with no dev license fees. That would make Windows much better, but MS would still not in any way be required to open source (true GPL) the whole thing.
It should have been released under the BSD license and still should be.
Maybe the Illumos project will fix that by using the BSD license.
BTW, it’s nonsense to think the Solaris folks should work on Linux. Why not ask the BSD community to drop what they’re doing and contribute to Linux? … or any of the many other open-this-open-that communities.
I have never responded to a LinuxMag article until now. Joe, I think your comments that OpenSolaris developers should abandon OpenSolaris and work on Linux is way off base, and I agree with many of the posters that there is room for IllumOS in the pool. To paraphrase Mark Twain, “The reports of OpenSolaris’s are greatly exaggerated”.
Your comment that “Linux Won” also irks me. Why? It seems that more than one article I’ve seen in LinuxMag seems to refer to a war between OS’s. The idea that if it’s not Linux then it’s not worth working on is silly. Don’t get me wrong – I think Linux is the greatest thing since sliced bread having used Kubuntu, Slackware, Redhat, Suse, and others. But I am, by trade, a Solaris system administrator, and, as another person mentioned, you can’t beat uptimes of Solaris systems – some of our systems have been up for years. I also use and very much like OpenSolaris and I would hope that Oracle would come to its senses, but outside of that, I wish the IllumOS developers much success in the future.
Well, there does not seem to have been much activity on this since 2010, and I’m happy to see that the Illumos community is not only more vibrant and active than ever, it’s expanding very quickly. There are a growing number of great distributions with significant corporate backing, and the future of Illumos looks rosy and bright.
You say “Linux won”… You’ll probably also remember a time when people said that about Windows. I think we’d all be a whole lot better off if we focused on the fact that UNIX won, and remember that it did so not in spite of, but because of its inherent diversity.