Quote:
Originally Posted by devilsadvocate
MVS...old-school IBM virtual filesystem?
|
More than just filesystem: it was IBM's top of the line mainframe OS. MVS and UNIX were polar opposites in the way the file system was implemented. UNIX just saw files in a hierarchal file system, and didn't especially care what they were - that was the job of whatever program used the file. MVS
did care what a file was, and knew a good deal about the files and their contents and uses. So you got things like ISAM files (Indexed Sequential Access and Maintenance) files suitable for database use, and PDS (Partitioned Data Set) files whose members were what we might think of as individual files.
Despite mainframe origins, things like ISAM are still widely used. Until recent versions, for example, the ISAM file type (as MyISAM) was the default database file type used by MySQL. (The current default appears to be InnoDB, which looks a lot closer to a true relational database than ISAM is.)
Quote:
Hmmm...Still runs on z/OS; new versions of it appear to recognize the forward-slash path syntax of *nix-based filesystems (which would make sense given z/OS is a UNIX implementation). I look at that now compared to my own 6TB RAID-10 array (formatted XFS) and am reminded of Jevon's Paradox, wherein technological progress which increases the efficiency of resource actually increases the rate of consumption of that resource rather than decreasing it (which was ostensibly the goal of the innovation in the first place).
|
z/OS is what IBM calls MVS these days, and is a lineal descendant. It's not quite a UNIX implementation: it's MVS with a POSIX compatible subsystem that looks and acts like UNIX, and is partially integrated with the main system. (The POSIX spec is based on UNIX.)
Years back, AT&T was still in the computer business, and entered into a joint venture with Sun Microsystems to create a converged version of UNIX that would marry the best parts of the AT&T System V and BSD flavors, under the direction of Sun R&D VP Bill Joy, who had been chief architect of BSD back when he was a grad student at UC Berkeley. UNIX licensees IBM and HP feared that the AT&T/Sun alliance would give them first look at the new code and a leg up in the UNIX marketplace, so AT&T spun off the UNIX division as separate entity called UNIX System Laboratories.
USL was later purchased by Novell when Ray Noorda was running the shop, but Novell never seemed to have a clear idea of what to do with it. When Noorda was removed as CEO, Novell sold USL to the Santa Cruz Operation. But they transferred the UNIX
trademark to the X-Open Group. X-Open published a set of specifications called "Spec 1170" that defined what a UNIX system was, and any OS that could pass those tests was entitled to call itself UNIX, regardless of its origin. (I recall a no longer extant Linux distro actually spent the time and money to pass Spec 1170 and get that right.)
It was theoretically possible for IBM to modify MVS so that it would pass Spec 1170, and could be called UNIX. I was deeply amused when IBM
did (though their goal was POSIX compatibility to make the Federal Government which was requiring POSIX in systems it purchased happy.)
______
Dennis