But the VAX-philopsophy was extreme CISC, and that
extended to the
VAX/VMS filesystem as well: record-based, with a zillion different
file
attributes, built-in file versioning, etc. It was hierarchial, but
mixed with a clunky "volume" concept (something like DOS "C:", but
with
longer names).
I wouldn't call it "clunky". In fact with VMS you can combine drives
into a volume set if you want the filesystem spread across multiple
drives, or you can treat each drive as an isolated filesystem. As a
system admin, I prefer the separate volumes, makes it easier to manage
overnight backups. In our VMScluster individual disks are backed up in
parallel across several tape drives (as many as four tapes in operation
at the same time, depending on day of week), a trivial task in VMS but a
bit more elaborate on a Unix system.
As for the RMS file attributes and versioning, they are a dream for
programmers, compared to PC or Unix systems. We have NT or W95
workstations at every desk, but we still keep a VMScluster running,
partly for financial apps, and partly for coding. When you start
dealing with larger apps (i.e 3K-5K users per week, 300-500 at any one
time in a 24 hour day, all accessing the same files) you start to
appreciate what VMS can do.
BTW, one nice advantage of filesystem per disk is drive shadowing in
VMS. You can mirror two drives during the day, then break the set,
remount the mirrored drive as a separate disk, back it up, then
reconnect it back to the shadow set. Instant snapshot backup without
shutting down applications, all the database files are intact as of the
moment you broke the shadow set, no updates during the backup.
Jack Peacock