Page 1 of 1

Things that changed from MVS to z/OS

PostPosted: Mon Jul 26, 2010 1:56 am
by amhajja
Hello all,

I am trying to write a research paper about a couple of new aspects that changed greatly between MVS and the implementation of z/OS.

I would be grateful if you could give me some ideas and resources on where I can read about them.

Thank you.

Zoro

Re: Things that changed from MVS to z/OS

PostPosted: Mon Jul 26, 2010 2:26 am
by Robert Sample
First, you cannot forget that between MVS and z/OS there was OS/390. You might want to consider your question a bit more carefully, too. The first incarnation of MVS came out about 1974, the final version (MVS/ESA) dates to roughly 1988 and OS/390 in 1995. So there's 21 years of MVS to consider. Comparing the 1974 version of MVS to z/OS directly will uncover many, many, many differences -- but many of them were actually part of later releases of MVS and cannot be attributed to z/OS at all.

Just for one example, when I started full-time employment in 1977, Amdahl was attempting to install a fourth megabyte of memory on our 370/168 running MVS -- something IBM said could not be done but Amdahl said could be (and eventually managed to get working, too).

And Google is your friend. Googling mvs history returns 618,000 hits so I'm sure you can find all sorts of good material.

Re: Things that changed from MVS to z/OS

PostPosted: Mon Jul 26, 2010 7:34 am
by steve-myers
The other problem is "MVS" is very much a moving target. You have to remember that OS/360, the direct ancestor of MVS was a strange mixture of 16-bit code (some critical data areas had to be located in the first 64K of storage), 24-bit (16-meg), and 31-bit/32-bit (depending on whom you talk to) code. The 16-bit stuff was not expunged until MVS/XA. The 24-bit code is still with us; just look at the DCB and DEB data areas, and I'm sure a whole host of other data areas. The differences between the very first MVS in about 1974 or 1975 and the MVS most of us actually started to use in the later 1970s were quite substantial.

In 1975 (if I remember correctly) my employer shipped several of us to Chicago for a two week MVS course. One of the exercises, which we skipped, was trying to tune the system using the early prerequisite to the tuning facilities in later MVS releases. We skipped this because by the time we went to Chicago it was obvious our shop was not going to this MVS. One group worked very hard on this exercise, and their report was very discouraging. It seemed to them that all of the tuning tricks they tried seemed to work the exact reverse of what they thought they would do! In other words, MVS was quite uncontrollable!

We eventually did go to MVS, in 1977 if I remember correctly. The big reason was to utilize the 370/168 attached processor. The 168AP was a sort of crippled 168MP. It had a second 168, but the second 168 had no storage or channels. The second 168 shared storage with the first 168, and all I/O was done by the first 168. In any event, we brought the system up on a Monday morning, with a strict dictate from upper management that we would return to OS/VS2 Release 1 when the MVS crashed. Monday came and went, Tuesday came and went. Wednesday, upper management threw in the towel and told us to leave it up. By that time VS2 Release 1 would have crashed several times, so we were way ahead!

Re: Things that changed from MVS to z/OS

PostPosted: Mon Jul 26, 2010 9:15 am
by steve-myers
Actually, now that I think about it, the 16-bit OS/360 control blocks actually had to be in the first 32K of storage. This changed late in the pre-MVS/XA releases to the first 64K, and this change broke some programs, though the repaired programs would run perfectly well in OS/360 if they did not have other MVS dependencies.

Re: Things that changed from MVS to z/OS

PostPosted: Mon Jul 26, 2010 10:11 am
by dick scherrer
Hello,

Also, one might look here for some history:
http://en.wikipedia.org/wiki/History_of ... ng_systems