Slashdot Mirror


OS/390 Replaced By z/OS

n7lyg writes: "ZDnet reports that IBM is replacing the venerable OS/390 with something called z/OS. What I want to know is if using z/OS is still like 'kicking a dead whale down the beach,' as Ken Thompson once said of one of its predecessors (DOS? OS/360? I forget the exact OS he was complaining about)." Well, z/OS does add 64-bit support and other goodies.

22 of 215 comments (clear)

  1. DOS? by Anonymous Coward · · Score: 4

    I was under the impression that DOS is on the way out, now that Windows XP has been released. Why is IBM still basing a mainframe OS on it? Are mainframe programs _really_ that hard to port to a modern language like C? All that RPG and Rexx is so hard to maintain it would be cheaper to rewrite in a modern language.

    1. Re:DOS? by Anonymous Coward · · Score: 4

      All mainframe OSs have historically been based on DOS. It set the standard in 1981, when mainframes were invented, and changing now would break that, and numerous mission critical apps, as well. Ken Thompson doesn't know what he's talking about.

    2. Re:DOS? by supersnail · · Score: 3

      DOS ==> Disk Operating System.

      This was released sometime in the late 60s for smaller IBM mainframes.

      The current ancester of DOS is called VMS.

      This is one of several operating systems called DOS, from, several manufacturers and "VMS" is also the name of a completely different OS from DEC (now Compaq).

      Mainframe DOS has/had nothing in common with PC DOS, and, apart from running on the same harware it has nothing in common with "OS" which was the ultimate ancestor of OS/390.

      This is probably the most re-branded OS the world has or will ever see SOME of the names it has had over the years are "OS" "OS/MFT" "OS/MVT" "OS/VS" "OS/VSE" "XA" "OS/390" and now "zOS".

      --
      Old COBOL programmers never die. They just code in C.
  2. Re:Who cares? by hawk · · Score: 4
    > man, does everyone have to dis marketing people? They serve an
    > important part of any business


    They could also serve as a valuable source of emergency protein during famines.


    :)


    hawk

  3. Re:It was TSO by ninjaz · · Score: 5
    It was SC Johnson -

    http://nicse.hlrz.kfa-juelich.de/~cbest/postdoc.ht ml

    There are also legends about how TSO sure is hard to use, but it's slow...

  4. Re:Why not OS/360? by Guy+Harris · · Score: 3
    the "full circle of applications"

    ...meaning both number-crunching and business applications; the S/360 instruction set was, at its core, binary, but, in addition to the floating-point instructions, it also included decimal-arithmetic and character-string instructions. (The floating-point and decimal instructions were, I think, extra-cost options on the original S/360 models or, at least, on the lower-end models.)

    The next version was OS/370

    ...because it ran on System/370. I suspect they might've bumped the system number because I think the first S/370's came out in, err, umm, the 1970's; they did make some instruction-set architecture changes, although the big change was the addition of an MMU (which was also in the System 360/67, but that wasn't a "mainstream" S/360). The very first S/370's didn't have the MMU, but later ones had it, and there were add-ons for those other models.

    then came OS/380

    Not as far as I know. The "S/380" machines - which, I think, came out in the '80's - were, I think, given names like "3081" and "4300", i.e. they were sold with just model numbers, rather than as "System/380's", and the OS was probably just being called "MVS" at the time. (I think MVS may originally have been called "OS/370 VS2 with Multiple Virtual Spaces", to distinguish it from the version of VS2 with a Single Virtual Space - I think SVS just ran all processes in the same virtual address space, using the protection keys to keep them from stomping on each other, and using base-register relocation to allow programs to be run from wherever they happened to be loaded, just as was done on the non-virtual-memory OS/360, whereas MVS gave each process its own virtual address space.)

    and finally OS/390

    No prizes for guessing in what decade that came out. It ran on System/390's.

    Unfortunately, that numbering scheme has, err, umm, a bit of a Y2K problem - would this decade's machines be "System/3100"s, or "System/3A0's", or "System/400"s (which would run the risk of confusion with AS/400's), or what?

    So I guess they decided, now that they've introduced a 64-bit version of the architecture (not bad for an instruction set architecture whose design started in the early '60's, assuming it didn't start at the end of the '50's...), to come up with an Exciting New Name; I don't know whether that inspired this whole new "[a-z]Series" naming scheme, or not.

  5. Re:Unadorned TSO did indeed suck... by Guy+Harris · · Score: 3
    Depressingly like filling out forms in a web browser...

    About 5-7 years ago, I read some magazine (I forget which one it was) that referred to Web stuff (or, at least, the forms-based version) as "3270 for the '90's". The host sends a form to the terminal, with fields to be filled in. The user at the terminal fills the forms in, and hits Transmit (or whatever it's called; perhaps they even called it "Submit" :-)), and the contents of the fields are sent back to the host. The host then processes the transaction, and sends another screenful back to the terminal. Sounds familiar....

  6. Re:Who cares? by Ed+Avis · · Score: 5

    I don't think IBM's customers will like the new name. It contains one of those commie lowercase letters, and fails to include a dull-sounding number after the slash. How do they expect to get enterprise-class reliability with a name like that?

    OS/380 might have been more appropriate - though even that is a bit racy.

    --
    -- Ed Avis ed@membled.com
  7. Re:COBOL.net too! by Mike+Buddha · · Score: 4

    Don't forget object-oriented, net-enabled COBOL to go along with it.

    Ugh, talking about kicking a dead whale down the beach...

    --
    by Mike Buddha -- Someday the mountain might get him, but the law never will.
  8. Re:power computing by jms · · Score: 3

    People somehow think that open source was magically invented a few years ago. From the 1960s through the 1990s, IBM supplied the COMPLETE source code to their operating system, compilers, libraries, boot loaders, EVERYTHING, to their customers.

    The reason that a company would spend millions of dollars on a mainframe was not only because the hardware was faster. It was because the software was completely customizable.

    While the Unix community, as it was, was struggling with incompatable binary releases, bugs they couldn't fix, vendors who came and went, leaving object code wreckage in their wake, mainframe programmers had organizations like SHARE, specifically designed to allow people to trade and share mainframe source code.

    It wasn't truly open source, because the source code was only available to those organizations who had purchased IBM hardware, but only IBM hardware would run IBM software, so for all practical purposes, it WAS an open community.

    IBM had a disasterous change of management in the early 1990s, and some idiot made an "executive decision" that IBM would stop releasing their source code. This was a move that effectively destroyed the OS/390 and VM codebase, and caused customers to leave IBM in droves. Because customers were no longer to be trusted to modify their own system source code, IBM threw thousands of programmers at the source code, trying to make it everything to everyone, so that no one would ever need to modify anything. The result is that VM is no longer a lean, mean operating system. Instead, it's a bloated mess, filled with poorly thought out interfaces that very few people even use.

    So, in essence, you have it backwards. One of the main reasons that IBM was so profitable in the 70s-90s was because they DID make the source code available for "geeks" to toy with -- except that the geeks had titles like "Systems Programmer" and "Systems Analyst". And most of the major mainframe applications -- air traffic, banking, insurance, finance, were written in house, by those same "geeks" who were only able to write the software that powered the mainframe generation because they had access to the source code and the ability to modify it.

  9. Re:Who cares? by finkployd · · Score: 3

    The engineers probably still call it OS/390, and dutifully ignore the marketdroids, as we should too.

    Actually, a lot of us still call it MVS :)

    Finkployd

  10. Major differences by FJ · · Score: 5

    This was published by IBM some months ago, but this matches actual release of z/OS. From my reading on this there are a few differences between OS/390 and it's successor z/OS. Nothing earth shattering but some cool stuff.

    1.) Software cost. On traditional mainframes the amount you paid depended upon the size of the machine. The larger the box the more the cost. To combat this most companies run there mainframes at a very high CPU usage. I've been at places where 100% usage on the CPUs are not at all uncommon. I believe with z/OS this will change and you will be able to purchase based upon usage of the software. This makes it a much more affordable for things like internet business where sudden spikes in usage cause problems. No you can buy a huge box but pay for a small CPU usage. If other mainframe software vendors (like CA) follow this lead it could be a huge reduction in software cost for businesses.

    2.) CPU clustering. I believe OS/390 could only have a maximum of 16 CPUs. I've heard that z/OS can have up to 64.

    3.) IBM is going trying to force the n-3 upgrade restriction again. Basically IBM releases a new version of the OS every 6 months. If you keep current on maintenance they will assist in an upgrade from any version 3 levels old. For example, if you are going to OS/390 R10 they will help if your current version is R7 or later. IBM always pushed this but Y2K made them make this a strong recomendation rather than a requirement. This is still going to be a tough sell because most companies don't like doing OS upgrades every 2 years.

    4.) z/OS is designed to work on there zSeries processors to use all of the new features. I believe you can still run it on a later 9672 machine but you can't do all of the new stuff. That means most companies are going to be forced to buy new hardware and with Amdahl & Hitachi no longer selling IBM compatible mainframes, Big Blue is in a nice position. OS/390 R9 is the last release which will run on the old water-cooled bi-polar boxes. Everything now needs CMOS technology.

    6.) 64 bit address spaces have improved. 64 bit was introduced in OS/390 R10 (if you run a new zSeries machine). This really isn't a huge deal to me. The biggest advantage will be DB2 and other databases who will no longer need to use hyperspace to store there data. The average program will probably never need 64 bit addressability (some don't need the current 31 bit addressability). I think the main benefit of the zSeries machine's support of 64 bit is for non-z/OS operating systems (like Linux). The really wierd thing is that you can have one LPAR running 31 bit and another on the same box running 64 bit. It's one line in your LOADxx member to switch back and forth.

    7.) TCP/IP got another overhaul to, among other things, make it faster to communicate across LPARS on the same box.

    8.) In either z/OS or OS/390 R10 Work Load Manager gained the ability to manage resources across different LPARS instead of just managing an individual LPAR. For those of you who don't know what Work Load Manager (WLM) is, it is a cool little tool that allows you to define what your business goals are and it will manage CPU consumption accordingly. For example, if you can say that you want 90% of a specific transaction to complete within 1 second and 99% of them to complete within 5 seconds. WLM will then increase & decrease CPU accordingly to meet your goals. If it can't meet the goals it will report saying that it's time for either a new box of for you to get realistic about your goals. Its a really nice tool.

    There are others, but these struck me as the biggies.

    Later...FJ

  11. Re:Who cares? by QuantumG · · Score: 3

    man, does everyone have to dis marketing people? They serve an important part of any business. If it wasn't for them marketing people who you think do nothing your company would not have any customers and you wouldn't have a job. That said, most the good looking chicks in any tech company are in marketing, another good reason to keep the anti-marketing-ism to a whisper.

    --
    How we know is more important than what we know.
  12. Re:Who cares? by 1010011010 · · Score: 5

    If it wasn't for them marketing people who you think do nothing your company would not have any customers and you wouldn't have a job

    Ah, yes, the sales people are useful. But no one ever accuses Marketing of "doing nothing." It's usually accusations of doing too much of the wrong thing. Most of the boneheaded things that a company does seem to come from the marketing department. There seems to be a lot of hubris in most marketing departments.

    It's not as if "Marketing" hasn't earned their reputation over and over again by rushing product, lying about product, spending a zillion dollars on 6-color glossy ads for unfinished products while engineering lacks resources to finish them, calling up competitors and alerting them early to the company's new product, etc.

    But Hey! They're beautiful people making beautiful things, and they know better than the engineers and finance people what should be going on.

    I know of a certain marketing department for an ISP that designed the company website to work only with IE5.5+ and Netscape 6 with flash plugins. The whole thing was basically a powerpoint presentation. Fixed layout, annoying navigation, the whole package. When the engineers pointed out that they were effectively shutting out 20% of the company's customer base, and probably annoying the rest, the response was, simply, "we don't care; the website is for CEOs. They all have Windows with IE and want presentations."

    Engineering produced a version of the website's HTML that was also backwards compatible in an afternoon, and forced marketing to use it.

    Marketing should not run the show.

    - - - - -

    --
    Napster-to-go says "Fill and refill your compatible MP3 player", which is a lie. It's not MP3. It's WMA with DRM.
  13. Not a huge difference by Viv · · Score: 5

    I just had an IBM full-line presentation at work the other day, and the new 'z/OS' is just an evolutionary difference, kind of like just a new version of OS390.

    Basically, this whole xSeries/zSeries/pSeries/etc is just a new name, it's not a real change in the product.

    xSeries -- 'x' stands for 'x-architecture' which is their name for technology they've migrated from their mainframe line to increase the availability of Intel Servers. An example would be their 'lightpath' technology, which is quite nifty. Press a button, and an LED lights up next to failing components on the motherboard.
    pSeries -- 'p' stands for 'performance.' Just a renamed RS/6000.
    iSeries -- 'i' stands for 'integrated'. Same as the old AS400.
    zSeries -- 'z' stands for 'zero downtime'. Same as the old S/390.

    They're really just changing the naming. Why? Well, a good deal of it is that they want to differentiate the products and start getting some cooperation between server divisions instead of having them compete as much as they have in the past. The nomenclature very specifically positions each product in a specific market segment. x

    Series is for when you need the efficiency of an Intel server but the high availability features of a mainframe. pSeries is when you need performance. Need solutions in a box? Go for the iSeries. Can't stand ANY downtime at all, ever? Go with zSeries.

  14. In keeping with IBM naming scheme by Pseudonym · · Score: 3

    OS/390 was the OS for the IBM/390, and OS/2 was meant to be the OS for the PS/2. So the z/OS name is at least consistent.

    --
    sub f{($f)=@_;print"$f(q{$f});";}f(q{sub f{($f)=@_;print"$f(q{$f});";}f});
  15. It was TSO by Mr.+Protocol · · Score: 3

    "Using TSO is like kicking a dead whale down the beach," said ... I forget who. It wasn't Ken Thompson, though. It was another member of the lab staff. Anyone remember who it was?

  16. Meet David Boyes by MonkeyMagic · · Score: 4

    There's an interesting article on Linux Planet interviewing David Boyes (the man who had 41,400 Linux images running on a single mainframe using System/390.


    DILBERT: But what about my poem?

  17. Z/Architecture Documentation by SteveMu · · Score: 5
    z/OS runs on z/Architecture, which is documented in the Princples of Operation manuals at this site.

    More details on z/OS here.

    This is what Linux wants to be when it grows up :)

    Good bedtime reading !

  18. Re:Who cares? by SubtleNuance · · Score: 3

    man, does everyone have to dis marketing people? They serve an important part of any business. If it wasn't for them marketing people who you think do nothing your company would not have any customers and you wouldn't have a job.

    Lets take all the useless lies, propaganda and VanillaBS(TM)®© spewed by Marketing/Advertising people and liberate their time for more useful efforts. Those useful efforts may be childcare, eldercare, reading to children - or other real value efforts.

    I look around at the human/material waste associated with 'advertising' and am sickened... after 2000 years of recorded history we have to create an entire profession of liars to corrupt reality with lies from capitalists.

    Why not free the marketing people and allow them to become artists and teachers? Why the hell do we need _another_ full page glossy telling us that "brand X is better than Brand Y because it is FJDKJSLKJ" - sheer utter useless crap.

    Why do Marketroids exist? Why are they enslaved to be mouthpieces of undemocratic multinationals? Read .sig - This is going to be the largest RIAA, MPAA, DMCA, WIPO protest since Seattle. The problem is a hollow democracy, instead of community, growth and democratic participation we get marketing - dong like it? See you in Quebec City.

  19. Who cares? by Will+The+Real+Bruce · · Score: 3

    IBM supports their OS for large-scale business rollouts, not for the consumer. This shouldn't make much of a difference in practice.

    Ooo, they call their new version something else. Like, say, Win2k instead of NT5, or "MacOS X" instead of Rhapsody.

    The engineers probably still call it OS/390, and dutifully ignore the marketdroids, as we should too.

  20. power computing by deran9ed · · Score: 4

    Operating systems such as z/OS fall in place where others aren't neccessarily useful. The OS is geared mainly (or it seems) for high powered computing, something along the lines like a huge mainframe, which OS' like Linux, or the BSD's cannot be trusted to support.

    Not to start a flamewar of any kind but there isn't a company I can think of who would dish out cash for some huge mainframe-like computer solely to let one of their geeks toy with, and install anything other than something proven (or semi-proven via marketing.)

    Sure it may be biased on the geek level to discriminate against other Unixes but the fact remains money talks, and the companies using this OS and the servers they run on would be insane to let it happen at this point especially when Linux in my opinion is in such a disarray of distro's. there are no standards on many things, etc.

    Take a look at Motorola, they're power computing comes in the form of QNX, ChorusOS, Onea OSE, Integrity, ThreadX, for many high powered stuff. Are these less of an OS than Linux or any other open source based OS out there because its not "geek chic?" Hell no.

    AntiSpam Info