Changing CentOS in mid-stream

When it was a case of a few big servers, it was at least possible to upgrade them every two years or whatever. Try doing that with dozens of smaller instances.

Sure, you can automate it, but a major reason to isolate everything in its own VM or container in the first place is to avoid breakage when making changes to other components. You don't want to have to break and fix each of them very often for major OS upgrades.