As a specific example of something that I'm sure is fixed by now, when I attempted to implement a simple openldap server, the user was created with the default security in place, which left the user shell unusable for the rest of the install script, and it promptly bugged out. This was a very easy workaround, and I had a series of fixes jammed into the build script, like using /bin/bash for init scripts, etc (yes, I know, but that's the downside of running vendor hardware stuff).
Anyways, yeah. So take that pain, the fix having to be fixed every few months, and the tendency to build another set of (conflicting) libraries to work with 3rd party stuff that essentially was made for RHEL, I just gave up. That was 12.04. Maybe it's been fixed? Who knows.
I will say this... Getting custom vm's cobbled together and rebuilt was incredibly simple with their python vm tools, but I've already learned the RHEL build systems and gone through the work of getting it all working and fitted into a reasonable toolset. The simple inertia of continuing to work with a RHEL system vs rejiggering (technical term) all of my custom stuff to work with a distro that, let's be honest, does little more than host the various bits of middleware, db, and whatnot that is not part of said distro, why would I bother to switch without some serious reason?
It's not pretty, and I'm sure there are plenty of technically sound reasons to make a different choice, but when it comes down to business sense, NOT taking the time to switch makes more sense than doing so if your current system works just fine.
There have been attempts at various points in various companies I've worked with to change off of RHEL, and that has led to a few systems of debian, a couple true ubuntu, maybe a suse or two, and hundreds of RHEL. So we convert the others by reason of ease of maintenance and swallow our pride of what's RIGHT vs what's efficient.
That's servers, mind you. They all run apache and a jvm and the dbas are always wanting to use their own db packages, so the main deciding technical factor is can they run, say, dell openmanage. I've had to forgo quite a few nice pieces of software, like openvpn, because of weird and completely nonsensical conflicts.
Enough about that. I will not recommend anyone run Cent or any of the RHEL variants as desktops because they run so far behind that little things like Chrome will gladly move on to better and incompatible libraries necessitating some hacks, or waiting it out, or doing a desktop upgrade right in the middle of a project sprint. I run ubuntu on my desktop because it's good enough that I can hack it to do what I need, and the folk that concentrate elsewhere (networking, dbas, developers) can easily run it and then when they have problems we're running the same toolset and I can help them out.
This is my reality. Your mileage may vary.