File versioning. This is bodged on in all xNixes by embedding version numbers into filenames, then writing a multi-tens-of-thousands-of-lines shell scripts to track all the versions, make loads of symlinks so programs can find them if they don't need specific revisions, occasionally add new versions and remap hundreds of links, occasionally prune old libraries, carefully manage umpteen paths to where different libraries are kept... and this is considered normal and nobody sees anything wrong with it, when in fact it's a vast ugly kludge of a hack, now fossilised as package managers and so on.
All because it's a braindead filesystem.
For system administration I use etckeeper these days. That serves nearly the same purpose. But for programming VMS was superior unless you are disciplined and commit frequently enough.
Sadly, this simplification was, IMHO, a step too far and a bad thing to remove. But Unix's namespace is much simpler, and doesn't really understand and incorporate extensions and things either. Which is a limitation but it makes it much simpler.
The fact that most Unix GUIs, from NeXTstep to KDE, put back in and use file extensions to associate files with applications shows that this is in fact important functionality.
Without extensions, having version numbering doesn't make as much sense.
But a Unix with file extensions and file numbers would be less like Unix than is, say, Plan 9 or Inferno.
An ordinary user app doesn't need to know. A file version control system does, though. Development environments need to know. IDEs and programmers' tools need to know.
Also VMS had their own issues when it came to it's filesystem(s).
OTOH, the point here is that if facilities like this were available, tools like Git -- VCSes -- could be far simpler and easier to use, because a core part of the functionality they add is built into the filesystem.
Disk space is less of an issue now. This is a server OS for big systems. Terabytes are cheap.
All filesystems have issues.
Remember, all software sucks: http://harmful.cat-v.org/software/
I used VMS on various types of VAXen in the mid 1980s and I liked it because of its systematically designed command line interpreter. A Unix shell is a complete mess compared to that.
I switched to Unix in the late 1980s and to Linux later and found it technically more powerful. Lightweight processes, pipes and many more. Is there anything in OpenVMS that Linux is fundamentally lacking?