I spent much of last week reviewing the ITIC survey on enterprise server platforms. IBM AIX won in virtually every major category, from uptime, to time to patch, in every type of problem report. It appears that IBM missed the memo that UNIX was dead. It would appear that Red Hat missed a memo as well because it trailed in virtually every area and its poor performance actually allowed Windows Server to generally fall within the Linux/UNIX range of statistics. This wasn't a Linux thing either because SuSe Linux, from Novell, was right on the heels of AIX, making me wonder why anyone would even use Red Hat if they are truly getting these results.
Let's talk about Windows Server, UNIX and Linux.
The ITIC Survey
The ITIC Survey measured a series of things ranging from downtime, to time to patch. It breaks down the results from Tier 1 to Tier 3 problems and then compares the results by group. In most sections, AIX won and Windows and Red Hat trailed. One major result was that while year over year, the results were mostly flat, Windows posted an improvement that was near vertical, suggesting that were this to continue, Windows Server could pass the field in one or two years. Improvements typically aren't linear, but Microsoft is clearly investing at a level that is paying measurable results. This survey touched 400 companies in 20 countries and focused on C-level executives. It was started in 2008; the report covers 2008 and 2009.
Linux Truly Is the Heir to UNIX
With the exception of Red Hat, the vast majority of Linux platforms intermixed very well with the UNIX platforms -- with Solaris trailing the UNIX pack. This probably at least partially helps explain why Sun is doing so poorly and why IBM seems to be the biggest beneficiary of folks running for their lives from Sun and Solaris. Though I'm really not aware of a big Solaris to AIX wave. But Linux performed strongly throughout, with Ubuntu Linux doing surprisingly well.
Another surprise was that this is the first time I've seen the MacOS show up in a survey like this. It did better than most of the Linux versions being tracked. I would figure that the MacOS would be about as far away from a great server operating system as you would get, but the survey results say otherwise. The MacOS did surprisingly well. If there was any doubt (and I doubt there was), Linux did perform in the UNIX pack and is proving to be a worthy heir to UNIX. Novell SuSe came out as the champ in this survey by a nice margin. Were I making a Linux choice, I'd sure factor that in. (In fact, SuSe from IBM might be the best of both worlds, given the AIX performance.)
Windows Server Improved but Showcasing Mistake
While most of the platforms had relatively small improvements year over year, Windows Server generally improved in the 50 percent range. This wasn't just for Windows Server 2008 but Windows Server 2003 as well, showcasing a huge commitment by Microsoft into making Windows Server very competitive. However, I also think it continues to showcase a mistake that Microsoft made in pooling the two code bases in their desktop and server operating systems. This pooling created two problems. The first is physical, which this survey is showcasing as being addressed, and the second one is of appearance. Folks still see Windows Server as a desktop OS that is trying to be a server OS and discount it accordingly.
Microsoft's server OS likely should have a distinct code base (and it is getting more so), and have a more distinct identity that is connected to what a server does, not to a concept of how you display desktop information. Microsoft has to sell against the name and heredity of its platform. I'll bet that creates a drag on revenue in the 10 percent to 20 percent range alone. This is important, given that its current quarter quarter was down sharply. These numbers also likely reflect that Windows tends to have higher penetration at departmental and work group levels, while UNIX and Linux are more likely to be higher scale, redundant, offerings. This would tend to shift the results in favor of UNIX and Linux, based on loads. If Microsoft really wanted to optimize its strategy against UNIX/Linux, it likely should have at least two server platforms, one focused on the traditional work group space and another that would position against UNIX and Linux clusters so that surveys like this would be more Apples to Apples.
It is amazing how often something we think of as dead seems to come back. Then, like the old Monty Python movie "Holy Grail" and the "bringing out the dead" scene, it not only isn't dead but it seems to be feeling more alive every moment. This was true of the mainframe and now it appears to be true of AIX, which did surprisingly well. It is also fascinating how often things come in from left field like the MacOS, which is UNIX-based, and provide a result that we should have anticipated but did not. In any case, this was a fascinating survey to review. If you want full details of the results, contact Laura DiDio of ITIC.