Other comments have touched on most of the reasons why Ubuntu is a good choice, or why it would be a bad choice in some situations.
For me, it was the OS I was used to. And, as I've had to deploy a few CentOS and OES servers as well, I much prefer how Ubuntu/Debian configures things.
Apache, PHP, networking, cron, etc. All much easier to configure and harden on Ubuntu than on CentOS. Only thing I've found CentOS does better is starting and stopping iptables, and that's solved with a quick apt-get install iptables-persistent.
Most of this opinion comes from writing Ansible roles that work on both Debian and RedHat systems. Ubuntu was always easy to get right. CentOS always had some weird thing that required an annoying amount of work to work around. (Like it doesn't run Postfix smtpd in a chroot while Ubuntu does. Meaning I had to have different Postfix settings in master.cf on CentOS than I do on Debian.)
My first Linux was Redhat 5.1 (not RHEL), but I ended up switching to Debian Slink and OpenBSD for whatever reason. Since then I just personally find Redhat based distros a bit wierd or clunky (based on lack of familiarity), and prefer to stick with the Debian/Ubuntu side of the fence.
Reason to prefer Ubuntu LTS over Debian Stable: fixed 5yr support period instead of a variable period.
Reason to prefer Debian Stable over Ubuntu LTS: all packages in repo are in the same security patching regime. With Ubuntu you have to be a bit careful using packages from Universe or Metaverse.
I did not know Debian didn't have fixed lifetimes for Stable.
Though they do have some form of LTS. [0] I do find it odd that the Debian Security team doesn't manage it.
Part of my RedHat dislike is unfamiliarity. But when I sit down and think about why I prefer how Ubuntu does something vs. how CentOS does something, I usually find Ubuntu methods make more sense.
For example, how the default apache mods are enabled. Ubuntu has the mods-enabled directory. CentOS has the LoadModule lines and mod config in httpd.conf. So to disable an unneeded module, you have to find the LoadModule lines, remove them, and the find all the various default config that is now broken. In Ubuntu, you just remove the symlinks in mods-enabled.
A new Debian Stable comes out when it is ready - after a long freeze.
Originally (from memory) Debian stable was only supported until the next stable release which could be anywhere from 18-36 months later (release timeframes seem more consistent these days).
Then they started supporting an oldstable (the previous stable release) for an extra 1yr timeframe.
Debian LTS is a newer initiative, but it doesn't quite have the same security service level as say Debian Stable. Not all patches make it to LTS, and those that do arrive noticeably later. Debian is trying to find ways to improve this though.
For me, it was the OS I was used to. And, as I've had to deploy a few CentOS and OES servers as well, I much prefer how Ubuntu/Debian configures things.
Apache, PHP, networking, cron, etc. All much easier to configure and harden on Ubuntu than on CentOS. Only thing I've found CentOS does better is starting and stopping iptables, and that's solved with a quick apt-get install iptables-persistent.
Most of this opinion comes from writing Ansible roles that work on both Debian and RedHat systems. Ubuntu was always easy to get right. CentOS always had some weird thing that required an annoying amount of work to work around. (Like it doesn't run Postfix smtpd in a chroot while Ubuntu does. Meaning I had to have different Postfix settings in master.cf on CentOS than I do on Debian.)