The Illumos announcement, aired last August 3, may be the soap of this Summer. But it can also be the starting point of a story that will leave a memorable footprint in open source software. And that is not because of the project itself, which is cool. It is not because it shows some interesting advantages of open source software business models.
Above all (and if everything stays the same), we will remember this affair due to the huge opportunity Oracle is missing in this precise moment. Oracle is missing a healthy project, OpenSolaris, with a committed community of users and developers around the world, in one of the most critical market segments at this moment: high performance operating system platforms. What a blunder…
Some recap of the highlights before going on with my analysis. As you all know at this point, in 2009
Oracle bought Sun Microsystems paying $7.4 billion, just few weeks after IBM gave up negotiations for $7 billion. Oracle’s CEO Larry Ellison reportedly declared that Sun’s Java programming language was “the single most important software asset we have ever acquired”. Another landmark product of Sun, the Solaris operating system, focused on high performance computing centers, was also a main attraction to close the deal.
Over the following months, expectations created right after the announcement faded out, giving way for skepticism and uncertainty about the real intentions of Oracle in this operation. Further decisions in Oracle and subsequent news did not contribute to clear up the growing doubts.
Solaris 10, the latest stable version of the original Sun enterprise operating system, could be used for free, without any restrictions (including commercial purposes), if you did not want to get official support from Sun. However, Oracle decided to stop this practice as of April 2010. Now, you can download for free only a 90-day trial version (see Oracle’s new license agreement for Solaris 10, section 3 (a) ). The slogan “Solaris 10: no longer free as in free beer” became quite popular. It also left many corporate Solaris clients with a difficult decision: do we pay the license and new support service or shall we
move to another platform?. Millions of dollars in many companies are at stake.
In addition, Sun had started another thrilling project in March 2007, when they informed that Ian Murdock, founder of Debian, was hired to lead the so-called Project Indiana. This eventually led to OpenSolaris, the community edition of Solaris operating system. The first OpenSolaris release is dated May 5, 2008. Nevertheless, several months after Oracle’s acquisition something started to change. The OpenSolaris community formerly used to have fluent relationships
with Sun as their primary sponsor. Then, they had no clues about the official policy to be adopted by Oracle. And the red giant remained (and still remains) silent. No feedback, no comments. This forced an ultimatum to Oracle by the OGB (OpenSolaris Government Board). Oracle must appoint an official liason with the OpenSolaris community by August 16, 2010. Otherwise, they will return community control to Oracle (that is, they give it up).
Before this, 2 lead developers in OpenSolaris left Sun to join Nexenta: Garret D’amore and Richard Elling (expert in the powerful ZFS filesystem). The motivation behind these signings is NexentaStor, Nexenta’s main product focused on enterprise storage systems, based on OpenSolaris and ZFS.
The last chapter of this tech novel is the announcement of Illumos, performed by D’amore on August 3. Contrary to many comments, Illumos is neither a fork nor a new distribution. It is not a fork because Illumos will remain 100% compatible with OpenSolaris, thus remaining in the same development path. It is not a distribution, either, since Illumos will not release a packed product ready to be installed and deployed. Following the description by my colleague Miquel Vidal, one of the founders of Madrid OSuG, Illumos is a “code derivation”, aimed to create a 100% libre software branch from the main OpenSolaris source code. Thus, Illumos is a baseline product, only source code on which we can build our own distribution (like NexentaStor).
Now, my comments about Oracle’s undefined position and Nexenta’s move. First, many people think that Oracle is wasting the opportunity to take over the OpenSolaris community on purpose, due to some obscure, evil motivations against the open source movement, as a whole. I disagree. In my opinion, Oracle is a victim of an archaic business and outreach policy, maybe useful for private software but pointless in the open source arena. They avoid making any official comment about any Oracle product that has not been released yet. Including the next release of OpenSolaris (even though it is a community product not a private software commercial product).
Nothing can damage more an open source community than lack of transparency and feedback. Specially when the business model relies on a single-firm that sponsors the whole project (formerly Sun, now Oracle). Even better, Oracle shouldn’t make a great economical effort to maintain the OpenSolaris community at this point. The main initial push was already lead by Sun (in terms of funding and development time). This way, Oracle could have their own baseline community edition for their brand new operating system, just like RedHat has Fedora and Novel has OpenSuse. The many benefits in terms of cutting-edge innovations to be incorporated in Solaris and stability tests should not be underestimated.
On their side, Nexenta is driven by a similar interest: ensuring the sustainability of their NexentaStor solution, based on OpenSolaris. Unlike Oracle, however, they know what to do in an open source ecosystem. Instead of remaining silent, Nexenta decided to make the first move, giving the whole community a breath of fresh air. Besides, they make it clear that their intention is to ensure compatibility with OpenSolaris, and they invite Oracle to contribute. Yes, Nexenta invites Oracle, not the other way around.
Now, Oracle is missing the community lead. In a single-firm driven community, this is equal to losing control over the innovation cycle. Nexenta can now work to ensure that the main development effort will be focused on their own priority interests. The longer Oracle remains inhibited, the greater their influence loss in OpenSolaris community. Eventually, in case that Oracle persists in this attitude, they will end up without any innovation or development source to improve Solaris. In other words: they will have another standard private software product, with no strategic advantage over their competence. They will have to spend many more development resources on their own to try to obtain similar results. Just as Oracle has been always used to do in the past.
Some competitors like RedHat quickly realized about that. Of course, since they already have their own baseline community in place (Fedora), they’re not interested in getting in charge of OpenSolaris. Their strategy was to attract as many customers as possible while in an uncertain atmosphere, offering great deals for migration plans to RHEL. It is curious that Oracle seems to understand some of the benefits of open source software, because they started to offer a recompiled version of RedHat Linux since 2006, offering their own customer support service and certifying compatibility with their product catalog. Now, they let RedHat the opportunity to strike back.
Honestly, I’m not very confident about Oracle changing their strategy in the short term. They do have people who really understand the new boundary conditions imposed by open source (at least, many lucid minds imported after Sun acquisition). Unfortunately for Oracle, none of them seems to be right now in a top-level decision-making position in the company.
Therefore, we may remember this episode as the great Oracle’s gaffe. Or could they change their mind?
The “Battle for Sunlight” has just started. Let’s see what happens in the future.
*Update: Well, only one week after I wrote this post, Oracle finally made a decision about the future of Solaris 11 and OpenSolaris. Not suprisingly, the have decided to drop off the former community initiative started by Sun Microsystems in OpenSolaris. You can read the news on Slashdot about the announcement. The internal communicate also states that Oracle is aimed to focus primarily on the release and support of Solaris 11 to maintain it as a flagship enterprise operating system, resurrecting Solaris Express as a trial-only version. As I wrote in my original post, I don’t think this is a very smart decision. However, Oracle is too used to the “traditional way of doing things” as to be really convinced of the real benefits of having a FLOSS community to support your products with innovative features. This also paves the way for other companies to leverage the OpenSolaris source code. Who knows?
Maybe we can see an alternative enterprise OS, compatible with the original Solaris, competing in the market with Solaris 11. Of course, other question is if Oracle is willing to maintain compatibility in their “official” Solaris version. Time will tell. *