It is possible? Yes.
Has it be done? Yes.
Are there some issues? Of course (and of course you won't be told which they actually were, as who took the decision have all the interest in minimizing reports of issues and maximizing the news about savings, i.e. to report the success of the switch).
I'm well aware of that migration. I don't think they're really telling us the whole story (exactly like MS "factually" told us Vista was their best selling OS ever after its release), and keep in mind that it's pretty much a best-case scenario, and that they've been planning it since 2003 (9 years ago). They are well over time and budget (even though they're talking about savings now
). Of course, they say nothing about the percentage of Windows PCs left (still 20%?) or for what reasons, they haven't released documents that show their (likely creative) accounting (and if they're comparing to non-discounted licenses), how it affected user productivity, etc. But then again, it wasn't easy at all
, and they would likely have saved a lot too just by moving to a sane setup. I mean, they had one IT staff per 15 PCs (yes, really!), and their setup was pretty much a complete and unadulterated disaster. They had nowhere to go but up. No, it's not completely impossible for 100% of businesses but it's hardly like the completely clueless Ubuntu cheerleaders seem to think it is. You know the "just install Ubuntu instead of XP then just keep working as usual!" mentality which is completely detached from reality.
I worked with a company to do a migration from Windows NT 4 to Windows XP. The project lasted over 1 year and that actual "migration" period took about 6 months. There was a ton of testing with all their applications, particularly the database apps and the legacy terminal systems that were still in use.
Been there, done that
Well, pretty involved with the Win2k -> XP desktop migration at least, and moving servers from NT4 to 2003 (moving from NT domains to AD, etc). It was a major undertaking, costing millions and taking well over a year too. And we're talking about a migration between alike OS'es without too many differences (Win2k->XP is nothing like XP->Unbuntu for sure!) And while I wasn't part of the previous migration to Win2k, I can recall lots of problems we were having at that time, like applications all requiring admin rights (writing to folders regular users shouldn't be able to, having to track down the problem with filemon and regmon, and granting special permissions to users of each specific software through scripts) and similar things. I mean, it's already enough work and complex enough as it is...
We also recently finished our XP->7 migration at my current job, where lots of programs used on XP were Win98-era programs that were patched to barely run on XP and that just didn't work on 7. Like imagecraft 6, used for some old codebases where we only have to make minor changes occasionally. Or 64 bit drivers being unavailable for some necessary hardware... We had a whole lot of small issues to deal with, even if we were migrating to an OS that's very much compatible.
But if you're right about the practical obstacles involved, and then it turns out that Windows "9" isn't any better than 8, then it sounds like we're basically scr*wed in the long run. Would that be your assessment?
Pretty much, yes. But if Win8 turns out to be a complete disaster I don't see how MS could simple do nothing about it and aim for a even worse disaster. Do you really think they could have released a 2nd "Vista" (another problematic and poorly received OS) straight? And by the time Win 9 is out, lots more companies could have ported their software to run on OS X too, and maybe that the virtualization & terminal server-like solutions will improve significantly*. I don't know. It's hard to accurately predict the future of technology a decade ahead I guess. The batteries on my crystal ball are dead.
* Right now it mostly means moving your software from cheap "desktop" commodity hardware to enterprise-grade servers that cost tens of thousands of dollars (while still still being accessed from the exact same commodity desktop hardware, so no real savings there typically), then having to pay for the virtual instances of the OS that runs on it (many very expensive Windows server licenses to buy!), then user CALs (~$30/user), then having to buy terminal server CALs ($800 for 5 users) on top of that, then perhaps Citrix CALs too ($945 per 5 users for XenApp Fundamentals), assuming those server instances aren't running virtualized under something like vSphere (starts
at $995 per CPU), it typically requires some new networking hardware, perhaps an expensive upgrade to an existing SAN (a 5 or 6 digit amount of money), new backup software for those new servers (expensive Backup Exec and/or Veeam licenses) and a medium to backup to, often getting expensive paid support on a lot of things, a lot of planning being required by highly paid experts and so on. So yeah, it's definitely not a cheap option. And in the end, it's not as easy or responsive as just running the app itself on your PC. In my opinion, it just moves the "problem" somewhere far more expensive.