The end of OpenSolaris?


[updated following Mo’s comment]

Just five years after an even that sparked a global celebration, and one that I was proud to be a part of, the dream appears to be over.  And what promised to be the start of a new era, if not of dominance, at least of a renewed competitiveness, proved to be a chimera.  Yes, just five short years after Liverpool won the Champions League in Istanbul, the club’s constrained ambitions were fully revealed by the signing of the distinctly limited left-back Paul Konchesky.  Oh, and Oracle confirmed that the OpenSolaris project, launched in 2005, is effectively dead.

Adam Leventhal offers a rather interesting perspective on OpenSolaris, and it’s a perspective I share much – but not all – of.  But to understand how the project could make so much sense for Sun and so little for Oracle, I think it’s necessary to understand the mentality at Sun.   Specifically, the dot-com boom was a success from which Sun never truly recovered.  The company grew far more rapidly than seemed advisable, quality and culture the obvious victims.  And the aftermath of the dot-com boom, and the emergence of Linux and horizontal scaling, placed Sun in a very obvious bind.  Sun’s previously unassailable technological advantages were being attacked by that most dangerous of competitors, a commodity market.  Nothing stung (nor stuck) more than the label Merrill Lynch pinned on Sun in 2003, that is was in danger of becoming “irrelevant”.

The response, initially ungracious, turned to one of acceptance, respect and accommodation for what appeared to be a cyclical phenomenon.  Greg Papadoloulos, Sun’s CTO summed up the world view, as “Red Shift” (a term which refers to the property of light from astral bodies that are moving away from the viewer).  The idea was that the technology market was always subject to commoditisation,  a sort of Moore’s Law writ large: processors, storage, software – all would be subject to lower and lower prices (and it is staggering when you consider what 1GB of storage cost 10 years ago and what it costs today).  Sun’s strategy, then, was to look for the “red shift” segment – the segment of the market that want more than commoditisation will offer, that want to stay “ahead of the curve”, for whom technology is a prime competitive weapon, and a centre for investment.  This is actually a throw-back to the dot-com era: Sun was still hoping to capture the startups, and if Sun had capitalised on the the first boom, it had missed out badly on “Web 2.0″.

Sun aspired to be central to the next network-inspired boom, and it was understood that to do that, you had to be promiscuous, available, familiar, and easy-to-acquire.  Hence open source, and hence OpenSolaris, and hence creating an OpenSolaris distribution (rather than just offer the source).

The OpenSolaris distribution, project Indiana, was controversial for its two reasons – firstly, it exposed the fact that Sun retained the trademark “OpenSolaris” and could and would use it to further its strategy.  How serious was this?  Even if you felt that OpenSolaris did need a binary distribution, it was deeply divisive.  For example, it caused the esteemed and founding OpenSolaris Governance Board member Roy Fielding to resign from the project, commenting “this well is poisoned“.  This controversy also partially hid the second issue (something that Adam identifies, and which in truth predated project Indiana), that a lot of OpenSolaris development was aimed at making something usable on the desktop, rather than an enterprise-class operating system.

The question that was constantly aimed at Jonathan Schwartz and other Sun executives (and is one I still get, working as I still am, for an open source project), is “how do you make money?”.  Sun was always somewhat evasive on this point – but essentially Sun would continue to make money from servicing its enterprise customers and defend that business, while strategy was geared up to become a ubiquitous – i.e. highly relevant – vendor once more.

What does Oracle care of this? Not a great deal.  For all their red-blooded tough-guy image, Oracle are as much farmers as they are hunters, and they understand that it is much easier to rertain customers than to acquire them.  What does Oracle care for the red shift?  By definition, it’s a sector that doesn’t have much money, unlike Sun’s traditional customer base.

When Merrill Lynch called Sun “irrelevant”, it also went on to say that it expected Sun would “eventually be acquired for its installed base“.  That is debatable, because by the time Sun was acquired by Oracle, it was a veritable smörgåsbord of technologies that would have interested Oracle, including Java and MySQL.  But still, it’s an interesting lens to view the acquisition through.  Oracle cares much more about delivering an enterprise-class operating system than it does about providing useful and accessible technology for developers.

So, farewell OpenSolaris?  As the project is currently constituted, it seems so.  But happily, there is a community around the code, a community that has grown and diversified and is now genuinely independent.  The one aspect of the project that was most roundly criticised in 2005, the CDDL, would appear to be one of the things that has most guaranteed the longevity of the community post-Sun.  Participants in the Illumos community appear (to my non-lawyerly mind) to be safeguarded  from any claims from Oracle, at least for the code released under the CDDL.  Oracle’s aggression towards Google over Android is not a great concern for Illumos.

And what of Liverpool’s back four?  Well, with the arrival on Konchesy and Raul Meireles we might suppose that Roy Hodgson intends to play a 4-2-3-1 (pioneered by Benitez in Liverpool’s 2007 defeat by AC Milan, unless you know better, and almost universally adopted at the 2010 world cup).  What’s more, Liverpool have retained their crown jewels in Gerard and Torres, and we know that 4-2-3-1 is system they both shine in.  The future for OpenSolaris and Liverpool could be so much worse.

About these ads

10 Comments

Filed under Software

10 responses to “The end of OpenSolaris?

  1. Awesome analysis, but I’d like to make one salient point;

    For oracle, java and mysql mean you’ve got a thorn in the side of google, and it seems they are going to use it to poison them slowly.

  2. patrickfinch

    @Karl – are Google distributing anything with MySQL? You saw that HTC is paying Microsoft royalties on Android?

  3. Google are using mysql on a massive scale, not distributing it, but oracle can use their reliance on it to cause difficulties for them.

    I didn’t see the HTC/MS thing, but it hardly surprises me that it comes to this… Innovation is too difficult now because people are too scared of who owns what obscure broadly worded patent…

    *sigh* I wish someone would fix this world…

    • patrickfinch

      Not sure about MySQL at this stage – Google has the clout to fork, if they haven’t already. They certainly don’t need the trademark, and if Oracle has any patent leverage over Google for databases, it presumably had it before the acquisition.

      As for fixing the world – the problem is, you’d have to do a clean-room implementation :)

  4. Pingback: links for 2010-09-01 « Wild Webmink

  5. You make a ton of valid points, but my main question is that if Hodgson plays a 4-2-3-1, who are the two back midfielders? Poulsen? Is he a legitimate everyday starter?

    • patrickfinch

      Two out of the three of Poulsen, Lucas and Meireles, with Gerrard playing behind Torres. Even if Poulsen is a starter, don’t think Fantasy League for a second. He’ll get fewer points than Alonso did.

  6. Mo

    Your line about GPL vs CDDL is a little wide of the mark. The reason Google isn’t shielded from patent issues is because they wrote their own VM and *didn’t* use any of Sun/Oracle’s GPL’d code. If they had, they would have been fine (assuming the patents in question where implemented by upstream code, which isn’t guaranteed — it’s no accident that Java SE is GPL’d and Java ME isn’t).

    • patrickfinch

      You are of course right about Google and Java SE. There are two issues: 1. patent grant for derivative works under GPLv2, 2. patent grant for implementations conforming to Java specs. The grants under GPL v2 would have offered protection had Google not written their own VM, as you say, (although not as much protection as GPL v3, I believe).

  7. Limited software with limited hardware for mad money — for these, who ‘knows what they buy’. Right, DB is like that. New marketing way.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s