The university IT environment is unlike any other -- even the large enterprise serves a limited constituency by comparison. And the small university is more diverse, in terms of the needs and capabilities of its user population, than most companies.
So which OS is best? And what does that mean? By "best", do we mean:
Which OS offers the lowest TCO (total cost of ownership)?
Which OS makes the most efficient use of the hardware on which it is deployed?
Which OS allows the user to be most productive?
Which OS requires the least amount of training to maintain?
Which OS requires the least amount of training to use?
Well, the simple answer is YES, we mean all of these things. Each of these are questions which should be asked each time the IT organization seeks to acquire new hardware and software. But the answer we get will not be the same each time we ask the question -- either because the OS market has changed or the requirements of the project have changed, or the the user's needs or capabilities have changed.
I've already expounded, ad nauseam, about the importance of life-cycle funding so I just won't go there today. Instead, I will assume that the university IT department IS adequately funded to meet the recurring needs of the institution. (I know, this is NOT a foregone conclusion but it SHOULD BE -- even limited resources can be better spent on an ongoing basis than spent all at once with no plans for end-of-life replacement.)
It is time to dispel one widely-held notion -- that, in the long run, the selection of OS has much at all to do with TCO. In the end, it does not. Why? Because there are always trade-offs associated with our choices.
In the consumer marketplace this is not necessarily the case since the consumer does not value their own time and productivity as highly as their employers might. This is why I keep suggesting that educators supporting IT need to think a lot less like consumers and a lot more like CIOs.
Just as a for-instance, the annual cost of employing an IT professional well-versed in UNIX (or Linux -- the skill-sets are the same) is dramatically higher than the annual cost of employing an MSCE with a similar number of years of experience.
In this important respect the university has one strong advantage over the enterprise. Invariably, the enterprise has a workforce that is somewhat fixed. Its staff are all professionals with lots of experience but often not the time or propensity to learn something new. And since employee productivity is hard to measure in a department which is traditionally thought of as a cost-center and not a profit-center, management does not like to assume the costs of re-training its staff. For this reason, the enterprise might intentionally make less-desirable choices based upon the skills it already has in-house as opposed to the skills it has to go out and hire (or train.)
The university on the other hand tends to have a more transient workforce. They generally have less experience over all but they are younger and have the opportunity and the drive to learn new skills. In this environment, most of your staff are in-house trained and are more than willing to learn something new. As a result, the cost of staff plays a smaller role in the TCO decision.
So which OS is right? It depends upon the project. In a student lab for instance, the needs of the student are paramount. Will the student be doing their assignments using personal productivity software. Will they be dependent upon software which only runs on certain platforms? Will they be programming or learning IT-centric things in the site?
More likely than not the university will find that each of the three types of platforms will need to be represented. Windows, Macintosh, and UNIX/Linux. (Okay, we can separate off UNIX and Linux if we must -- but the environments are so similar that I am not sure why we would. )
In the university machine room, the choices are driven even less by the OS than by the need. If one needs scalability (tens of thousands of e-mail accounts for instance, or all student records, the choice may be based on the need for raw horsepower to handle large databases. If more subtle capabilities are required, there is likely to be more flexibility. But, if the need for compatibility with proprietary systems is the driving force, then there may be no flexibility whatsoever.
So what's my point? My point is that, in a university IT department, there is no room for bias for or against any one choice of OS -- or any one solution. Many factors must be considered and the needs of your constituents, whether they be students, research faculty, or the university administration, must be paramount.