It's an enterprise operating system. Nearly any feature you can think of that you would want in a large business is already built into the operating system. Clustering, applications built out of objects written in different languages, transparent network file support, automatic file versioning, backups, huge data sets, pervasive runtime performance and process introspection features, layered security, disaster recovery file set shadowing across the WAN, built in bdb2 like database, it's all there.

Unlike a typical Linux deployment, these operating system components were designed and tested as a whole, so the system tends to have little if any issues when dealing with unusual circumstances.

Enterprise deployments for things like factories, nuclear reactors, water plants, live cable TV stations, and 911 call centers require a level of stability that you can't get with things like RHEL or Windows.

I don't mean reliability, I mean stability. The Linux kernel is reliable, OpenVMS is reliable, but a typical Linux distribution is made up of a hundred components constantly churning on different release schedules, whereas it's normal to take a 20 year old Alpha binary written in 3 different languages and expect it to run on Itanium or x64 using an OS that's 2 major versions ahead.

Just like z/OS or NonStop, it's entirely reasonable to have deployments which never patch the underlying operating system, and even ones that do patch regularly do so on a far less frequent basis, with a well-defined procedure to canary the changes to avoid taking down production.