Do you have questions about the site? Write to bruce at perens.com .
List of all articles we've ever published.
Source Code
Developer's Documentation
RSS
XML Sitemap
XML News Sitemap
![[Valid RSS]](/images/valid-rss.png "Validate my RSS feed")
Sun's Big Gamble
It has been almost painful to watch Sun over the last few years. Scott McNealy, Sun's co-founder and Chairman, so desperately wanted to compete with Microsoft on the desktop that he often led his company in some very odd directions.
When you consider that Sun's strength has always been on the server, things like the purchase of StarOffice and the creation of the JDS are just plain baffling. Only when you put it in the perspective of "sticking it to Microsoft" does it begin to make sense.
During the dot-com bubble of the 90s, Sun flew quite high and made immense amounts of money. That money allowed them to explore all sorts of ventures that weren't part of their core business. It also allowed them to delude themselves into thinking that one day everyone would see the light and run the superior Solaris on their desktop, shunning Windows.
Yeah, right.
Sun only has three real product lines: hardware, Java and Solaris.
Believe it or not, their hardware is Open Source. The specifications for Sparc have been available for years, as long as you pay the licensing fee to Sparc International. Back in late 2005 Sun released the Verilog descriptions, accompanying verification suite and simulation models to their UltraSparc T1 (Niagara) multi-core processor. Of course, not everyone has a chip foundry at their disposal, but it is as open as you can get with a CPU.
Their successful suit of Microsoft, back in 2000-2001, directly led to Sun's realizing they could successfully GPL Java without worrying about a fork. Java is a brand, not software. If you can't pass the compatibility test, you can't call it Java. If you can't call it Java, it will be dead in the water. Sun controls the definition of Java. QED.
That just leaves Solaris. Solaris has always been a strong contender in the Unix market, and the latest version brings to the table some very advanced features such as ZFS, Containers and DTrace. Their official position has always been that Solaris is superior to Linux in every way, and while Linux may be suitable for limited use at the "edge" of the network, the core belonged to Solaris. If you were serious, you called Sun. Period.
Things change. Sun released Solaris X86, which ran on Intel-compatible processors as opposed to Sun's own Sparc. While they made lots of noise about supporting the x86 market, their actions spoke more loudly than words and their support of Solaris X86 was noticeably weak. They were accused several times of trying to kill off their own product, which survived only due to the small, but rabid fan base. The message was still there, that if you were pig-headed enough to want to run Unix on Intel, then they'd sell you something. When you got serious, come back and you'd have a good migration path to a real operating system.
But Solaris doesn't really make Sun any money. To compete with the perceived "free" cost of Linux, Sun reduced the license costs of Solaris to almost nothing. When someone purchases a $500,000 computer, plus a few years of dedicated support for another $200,000, tossing in the cost of the OS is a no-brainer.
Now, while Sun may have derided Linux for being a poor shadow of Solaris in the highest levels of the enterprise, there is no arguing that it has matured at an astounding rate. If you look at the feature sets graphed on a curve, it becomes obvious that Linux surpassing Solaris isn't an "if", but a "when". The features that Solaris boasts over Linux are more and more frequently appealing to a smaller and smaller niche at the very top of the market. Solaris may outshine Linux on systems with 64 or more CPUs, but just how big of a market is that?
It may have taken years, but I think Sun is finally realizing this. Someone finally looked at the numbers and realized the market for small, 1U & 2U servers is substantial, and while Solaris may kick butt on big iron, little iron is 95% of the market. As painful as it is to some people, the reality is Sun is a hardware and services company, not a software company.
So, where does that leave Sun and Solaris? Enter Linux.
By releasing Solaris under the GPL, Linux will be able to close the gap in record time. Sun is reported to be seriously considering opening Solaris under the GPL. With Sun's longtime attitude of superiority, I seriously doubt there will be much migration of code from Linux to Solaris. However, it means that Linux can integrate the best and brightest of Solaris. Give it a year or so and no one will be in a position to declare Solaris' superiority over Linux at all, core system or not. Any deficiencies that Linux has on big systems can be cannibalized and integrated. Sun can still sell hardware, but the R&D and expenses for the OS can be greatly reduced. The dedicated Solaris development team can be reduced as Solaris is absorbed and replaced by Linux, and the community takes over. Sun can finally focus on what they do best -- sell killer hardware and superior services and support.
Sun's recent certification of Ubuntu on their UltraSparc T1 Niagara systems only furthers this. Official blessing from Sun for Linux on their hottest hardware.
In the end Scott McNealy may actually have a chance at seeing Microsoft unseated as the desktop king, with Sun being a major contributing factor.
The big question is, can Sun pull the trigger? Opening Solaris under the GPL is signing its death warrant, no two ways about it. IBM has come to terms with the eventual replacement of AIX with Linux. Likewise HP with HP-UX/Tru64. Can Sun make that call and still be Sun?
What do you think? Will Sun take the final plunge and embrace the GPL for its (remaining) crown jewels? What will that mean for the future of Linux and of Sun?