Tuesday, March 30, 2010

Long Term Investment (or Solaris is Dying)

If you know me, you might be surprised to see me write "Solaris is Dying." Well, unfortunately I suspect that may be the case. Oracle seems to be making good on its promises to make its Sun acquisition profitable, but in the long run something is going to give.

Two things have recently happened that have made me think that Solaris may not be around in the next 10-20 years, as the next generation or two of computer engineers enters the workforce:
1. Solaris isn't so free anymore.
2. Changes to the scope of Sun Education programs.

While there may not be any DRM (yet) to enforce the non-free-ness of Solaris, the fact that the license only permits legal use of Solaris for 90 days without a support contract or other paid license and that security patches are no longer available to non-paying customers severely restricts access to all non-enterprise customers. If it doesn’t restrict access, it at least strongly discourages anyone who had an interest in trying out Solaris.

A Bit of History

Back in the old days (before Solaris 10’s 2005 release), Solaris had a separate paid license. Intel/x86 support was a joke and it pretty much only ran on SPARC, but since it shipped preinstalled on Sun systems which were the lion’s share of SPARC systems in the world, it was practically free if you had the hardware. All patches were free. When Solaris 10 came out, Sun changed the scheme. Solaris 10 would be free for anyone to download and use, and only support would cost extra. Work on decent x86 support also began in earnest. Patches, however, were restricted. They were no longer 100% free; only security and hardware patches would remain available to non-paying customers. Solaris update releases (also free), however, would include all the other patches, so if you needed the other bug fix or enhancement patches but did not have a support contract, you could just wait a few months for the next update release.

In addition to the Solaris side of things, Sun hardware was prevalent at many universities in the 1980s and 1990s, exposing students to Sun technology. The dates might be a little skewed as that was way before my time as a college student, but from what I've heard from others that were around then, that was roughly the era of Sun workstations. In any event, during that era people didn't have cell phones and students didn't have personal computers in their rooms, and the PC didn't really even take off in households until the early to mid-90s. Today most educational institutions run primarily Windows and Linux, and if Sun hardware or Solaris is present, it is very often as part of a legacy installation.

What Happened?

Here's my theory: the PC happened. And I don't mean PC as in PC vs. Mac, I mean PC as in Personal Computer.

With the PC came an increased accessibility to computers by the ordinary citizen. Microsoft came in with DOS and then Windows. IBM made an attempt at OSes with OS/2. And there was Linux, the closest thing to UNIX that could run on a PC. UNIX was still king in enterprises, but that would change eventually. The first computer that many people were exposed to was no longer a Sun or other UNIX workstation in college, it was a PC at the local electronics shop or at home. People got comfortable with Windows and Linux before UNIX systems ever came within reach, and before long, all three OSes could boast comparable features, at least comparable enough that the average computer user would not realize any significant difference.

The PC generation then entered the workforce knowing Windows and Linux, but not UNIX. There are two options for this new workforce: learn something new, or push what they are already comfortable with. Guess what happened?

The Old IBM Integrated Systems Strategy

Oracle says it wants to repeat the successful IBM strategy of integrated systems that put IBM in the “no one gets fired for buying IBM” position in the 70s and 80s, but I have to wonder if they considered why the IBM strategy was not sustainable. Sure, UNIX systems became more suitable for many environments than the original mainframe model, while IBM stuck with the mainframe, but I think it also had to do with accessibility. UNIX systems were smaller than the mainframe, and therefore more accessible to smaller companies. The PC, driven by the x86 processor architecture, was affordable by someone as “lowly” as the home user. UNIX replaced the mainframe, and then x86 replaced UNIX architectures. Each successor was cheaper to acquire than the previous as well, but I attribute that more to economy of scale due to the level of accessibility. Of course mainframes and UNIX remain around today, but in a significantly smaller capacity than they once existed. Of the UNIXes still here, Solaris seems to be the biggest in terms of widespread use. At least, besides my level of exposure to it, it seems the fact that when applications have support for Windows, Linux, Mac, Solaris, AIX, HP-UX, with support for AIX and HP-UX having been pulled many years before Solaris support was pulled would suggest Solaris as being more prevalent. (Matlab recently pulled support for Solaris/SPARC. Mathematica still supports Solaris but recently dropped support for other UNIX operating systems.) And I think that's related to Solaris having had a significant presence in educational institutions for quite some time.

Oracle has cut off Solaris from the hobbyist. The hobbyist will no longer even have an opportunity to know Solaris. And when the hobbyist becomes a college or graduate level professional, they will report to work not knowing Solaris. The market will shift away from Solaris as people choose to stick with what they are comfortable with. Just as Oracle seems to be focused on short-term profits, corporations in the future will be similarly focused, and using the existing knowledge base of the pool of potential hires is often much more cost-effective (in the short run) than training them on a new system which the employees have no existing experience with.

The Last Straw

Okay, so no hobbyist. But educational institutions? As far as I can tell, the Sun Academic Excellence Grant and Sun Education Essentials hardware deep discount programs have both been discontinued under Oracle, with mumblings that even the standard education discount may be significantly reduced or even eliminated. Why wouldn't they be? They aren't exactly programs that make an obvious profit. But by killing those two programs, accessibility to and interest in Sun technologies is significantly reduced at the university level. As I mentioned earlier, college and graduate students won't know about Sun or Solaris when entering the workforce. Maybe Sun didn’t know how to market its products or was unable to develop a profitable roadmap, but I think it at least got one thing right, and that was investing in the future by increasing accessibility to Solaris and promoting its technologies at education institutions.

Is Oracle trying to kill Solaris? Maybe it is. If so, it's certainly done a good job without making it blatantly obvious. It will be a slow death. Maybe something will happen in the OpenSolaris realm to keep the goodness of Solaris alive. Or maybe I'm all wrong and everything will turn out fine. But considering that Solaris already has a limited presence in the minds of students today, further restricting access to Solaris will not help that. And it is certainly the students today that will be the technical employees and business leaders of tomorrow.

Food For Thought
  • Sun Microsystems was rated the #13 best place to work in IT by Computerworld in 2009. Oracle did not make the list.

  • If you ask the average (or even above average) college student today, they’ve heard of Sun, if only by way of an introductory programming course in Java, but have never heard of Oracle.


Keywords: oracle, sun, solaris, unix, linux

No comments:

Post a Comment