In 2004 a Linux distribution called "Ubuntu" came out. It's made huge strides. It has become so popular that it's become synonymous with the word Linux - not in any real sense, just in the way it's perceived.
However, for what's probably the most popular Linux distribution out there at the moment, it has some rather major problems. The problems seem to arise, much to my dismay, from the fact that it's the community supporting it.
And the problem is this: You can not expect enterprise level support. When Canonical (Read: Mark Shuttleworth) makes some big decision regarding the future of the distribution, such as the use of grub2 (while still in beta I might add) or the change in interface, he doesn't put any emphasis on resources for things like change control. In fact, the wiki seems to be full of documentation that seems to state "you just don't need to know how to do that".
But then when you start delving a little deeper, you find that the problem isn't so much with the dictated changes but more the lack of support. Take for example the netbook interface. Hugely popular but now being replaced with Unity. I'm not sure how many people have had a look at Unity but personally, I find it awkward, counter-intuitive, confusing and intrusive. I found a bug report asking why the Netbook Interface is no longer supported.
It would seem that it is the same reason that gnometris was broken for so long (A bug if it was using a theme which just about every distribution at the time had as a default) and cheese is still broken (It just can't do video. Wxcam seems to be a suitable replacement but isn't quite as pretty). Lack of support.
In a community model you're reliant on someone being interested enough to fix it. So what happens when no one's all that interested? The project falls by the wayside. This isn't to say it's not still useful to people. It's just that it's not interesting enough for hackers to play with it.
I'm sure other distributions suffer the same problems. However, other distributions release when they're ready rather than forcing a release every 6 months. This 6 month cycle tends to happen with a great gnashing of teeth as things seem to have changed for the sake of changing and contain more bugs than a cockroach hotel.
Try filing a bug report. One of the first things you're asked is whether this is a problem with the latest version (usually not yet released). If it's not, they'll normally tell you you'll just have to upgrade.
But then, those in the know may ask about long term support releases. However, these are only really supported in the loosest of terms. I would have thought that it would have been paramount to continue to support the kernel for example - releasing new kernels to make sure that this version was current. This isn't the case.
I'm not saying that Ubuntu is bad. What I'm really saying is that it has holes. Perhaps Canonical need to find a model that works for having paid developers looking into supporting these dropped projects. If only to provide a little certainty for system implementers. There's bound to be a fair few of us out there trying to make it work for large'ish deployments.
Of course, I can't complain. It's not like I've paid any money for it. My sense of entitlement is more frustration that I'm needed. The secret about most computer people? We work with machines that are designed to do repetitive tasks. In which case, we're lazy. "Is there an easier way to do it? Sure - I'll get the computers to do it". A computer can be a bit of a hammer - every task starts to look like a nail. In other words, every task is a repetitive one.
So how do you use Ubuntu or any other Linux distributions in big deployments? Hire a Linux geek? Except Linux geeks cost a hell of a lot. I was looking after a very similar issue - testing the compatibility of applications on a new version of Windows. Except that job wasn't quite so strenuous (in the same way). Assuming that I was charged out to the client at around $200/hour (and getting about 1/5 of that myself), then you're probably looking at around $120/hour if you were to hire that person directly. Experience counts here.
Perhaps Ubuntu just isn't suitable for big deployments. But then, how many Linux distributions are? I hold high hopes for Red Hat Enterprise - except that it's horribly out of date and expensive. Others are trying to full this particular void - Xandros, Oracle Linux and Suse for example. Perhaps this is where the more embedded approach such as Google Chrome OS (I still hesitate to call it an OS) and Meego come into play.