In the early 80's I worked for a small loan finance company in Des Moines, IA. They did processing and printed statements for a large number of similar companies world wide. We would IPL a 370-145 every morning with ACP9/TPF to support some 30,000 teletype terminals processing loan payments out in their offices. (same machine would be IPL'd again after the business day to MVS to support record keeping and reporting. The shop also had a Tandem machine, Burroughs, and HP small mainframes. A 3031 supported the programmers, statement printing, record archival, etc.. It was fun to be a computer Operator back then, you felt like the center of the universe.
I worked for a Fortune 100 food processor 1975/6 and the mainframe would crash a couple times each day. They kept score on a board on the wall and staff would plan their coffee and lunch breaks around the machine crashing.
@@KameraShy thanks. Yes it was common in those days. And it didn’t matter as much back then because computing was mostly batch and so they could easily wait
Thank you for this video. It is the most comprehensive review I have ever seen. I was glad to see you included DOS/360 as when I told people I worked on DOS they assume I meant MS/DOS. They never heard of DOS/360. It was the right operating system for many users.
Dad worked for IBM. In the late 80s he had a dial in and I used VM a lot. I also wrote a REXX script to solve a problem for him and it ran in MVS. Having used both, 15 year old me certainly preferred VM to MVS :)
A few things I remember from my time with big blue... The Mass Storage Subsystem which was a disk equivalent of virtual memory emulating huge amount of disks using a robotic tape system. A baseball mitten was apparently an integral part of the robotics. I once had to fix a mainframe that the customer had a major complaint about - the System Engineer had configured and ordered it with the wrong colour panels. In the end I had to order the correct paint and send the panels to a car paint shop.
Thanks for the walk down memory lane. I started my career fresh out of high school on a RadioShack TRS-80 in 1977, and within a few months landed a job as an application programmer on a IBM 370/138 running DOS Rel 34 with CICS. It was an assembler only shop, but I learned a lot. I then took a job as a systems programmer for a DOS/VSE shop running a 4341. When DOS began to run out of gas, we upgraded to a 4381 and ran VM with multiple DOS guests. This set me up as a VM/DOS/VSE systems programmer where I changed jobs a couple of times. After several years of that work our shop upgraded to MVS, so we maintained multiple virtual DOS/VSE machines as well as standing up an MVS instance, while we migrated data and applications over to MVS. I stayed as an MVS systems programmer for about 15 years (until mid -80s), and have spent the years until now as a software engineer on Linux, Windows and occasionally some mainframe jobs. I’m still working as a software engineer, and focused on AI, Robotic process automation and system integration work. What an amazing field to work in and I feel very privileged to have had the opportunity to spend my life doing what I love.
Fun fact: Interactive Systems Corporation had a version of Unix for the IBM 360 named ix/360. Kodak (yes, the film company) bought it in 1988 and sold it to Sun in 1991.
This was very interesting and brought back a lot of memories. I worked for IBM during most of the referenced time (1970 to 2000). Started out in final test for 360/40 in Poughkeepsie. I was in the first class for training test techs on the 370/155 DAT box (version 13). Moved on to Field Engineering Division where I made a lot of money doing DAT box conversions (adding Dynamic Address Translation) to standard 370/155 aka 3155) over a weekend. In the mid 70's I went the software route servicing DOS/VS and it's successors. Yes, there were a lot of patches (PTFs - Program Temporary Fixes). Ended my career in late 1999 working in the VSE System Center (Endicott) building VSE systems (SIE - SoftwareXcel Installation Express) for VSE users. We had our own data center with a 9370 processor running VM/ESA on which we built and tested each customers VSE/ESA system (including POWER, CICS, ICCF, VTAM). Our group also hosted user group meeting around the country for VSE users where I spoke on the need to prepare for Y2K.
1980s I worked for ITEL/NAS installing "Add on Memory Frames" to 370/155s. Hundreds of wirewraps to get another half megebyte or so of mainstorage. Also over the weekend. It paid my mortgage I guess.
I once worked on a project to keep a data center limping along until it could be completely decommissioned. Mostly changing tapes and operating printers. I got to know the guy who ran the place in its heyday. It was a large bank. They ran all their CICS and ATM transactions, as well as payroll, trust funds, etc. etc. in 512K of RAM in the 70s. When they upgraded to 1M of RAM, it was days of heaven! They had the first shared ATM network around town in the entire country. All in the RAM of an XT! My admiration for IBM is boundless.
I worked on banking systems running on IBM mainframes from 1978 to 2015. Wouldn’t it be cool to expand the timeline to also include hardware names and models - processors, dasd, tape, card, printer, etc. well done Moshix, I learned a lot.
When I worked for big blue in the 80s I once was invited to go sailing by the Data Centre manager of my customer and the assistant data centre manager had a S/370 console mounted on his bathroom wall. I asked him why he hasn't made the lights flash and the reason was apparently that each bulb needed 2A to light up and there were literally hundreds of flashing lights.
Thanks for the memories. I started my career in IBM systems programming back in the tail end of the OS/360 days on a 360-40. That didn't last long as they upgraded to a 4331. I moved to 43xx-30xx families and specialized in VM and MVX. I did a ton of work as a consultant to EI DuPont in Delaware at the Experiment Station (interfacing a Cray X-MP with a 3081 used as a front-end IO processor), and the textile fibers, P&EP and Berg Electronics divisions. I also dabbled in DPCX on the 8100 family. Good times.
This video brought back a lot of good memories! I worked with MVS/SP to Z/OS as an operator from 1983 to ~2010. All in all it was a decent career. I still have occasional dreams of working with mainframes!
Programmed a 370/115 running DOS/VSE in Assembly, COBOL, FORTRAN, RPGII, and JCL at the beginning of my computer career. I even knew how to program 029 and 129 card punches.
FINALLY !! Thank you for this Video !! I waited for 242 Episodes for THIS episode. This should be Video M001 !! Next Video - The Basic Architecture of a IBM 360/370 Mainframe - CPU - IO CONTROLLER - BUS Systems - Terminal-, Disc- and Tape- Controller etc. etc. PLEASE !!! And maybe a hind where to find the right document to approach the basic understanding of a IBM 360/370 Mainframe. And No ... i know there a over a 1000 IBM documents on Bitsavers and the Internet-Archive ... i need a hind ... a roadmap what to read in what order.
I worked on DOS and DOS/VS systems as a sysprog from 1970 to 78. I installed CICS in 77 also VTAM and NCP. You know you are on the leading edge when IBM SE's ask you how you did it!! Happy Days
Fascinating, thank you for sharing that. I'm a child of '62, started learning about (taking an interest in) computers in the late '70s. Always had this impression a Mainframe was so much more powerful than that, so much faster... To the youngsters today, the trick to understanding a Mainframe, is the airline booking system, the hotel booking system, the credit card system.... banks, earlier versions of the stock market systems... It is a system that has more resources than most home user / hobbyist computers, it has availability that's great. They had and worked with virtualisation since before personal computers existed. I can remember being told in about '82 that they could defragment memory, the same way you defragment a HDD, and they had two separate memory banks, and you could defrag active memory into one and de-power the other, then power it up again, then power cycle the 1st bank. If a program has a memory leak, loses track of the memory, there's only one way to recover that memory, which is to power cycle it. So for mainframes, they invented this system of having 2 separate memory banks.... It was a lot quicker to do that than it was to power-cycle the whole machine. The thing with a mainframe, was it had (let's say) 8 CPU cores, but you could shut down one or two or four, and hot plug them. Same with memory, HDDs, network interfaces, power supplies.... In a single cabinet, you had a back-plane / spine, and as long as you kept one motherboard plugged into the backplane, and power on, the machine was technically still up and could still respond to flight bookings and hotel reservations and things. The goal was that you may well have some issue with one part of the system, but there was almost nothing short of a nuclear war, that could take the whole thing down.
in 90' I work with PL/1 I found one word misspelled in the manual and IBM send me certificate thanking for the correction, I could read the manual write complicated code and run it the way I have intended too; It never happened again later in any other hardware, software systems, that is why we have Google and Stock Overflow, (PS if you had access to the red books you were a God at that time, another famous "color" book was: Brown Blue book ;) jobs ref )
Just to comment on how unique IBM's mainframe offering was, I once reviewed a UK bank's attempt to use Windows to run even a part of their processing. Just getting connectivity from 10,000 plus user PCs to a single Windows server was a major headache and required them to write a lot of special code to achieve it - IBM's mainframe was definitely the most scalable technology of its time.
In early 70s we had access to VM/370 in high school. I liked virtualization. Someone told me that someone managed to boot VM/370 inside itself 9 levels deep! Then when X86 hardware virtualization came out, I had to make it boot inside itself 10 levels and run Linux inside that (s.l.o.w.l.y) :D It was fun knitting pagetables.
@@moshixmainframechannel I sort of remember it was at somewhere like carnegie mellon (50 years ago), may have been chopped-up to make it work. but whoever told me (I don't remember) could have been wrong/making it up. you have to present a virtual machine with all the features that the hypervisor uses to run virtual machines.
Thanks for the history. I worked with mainframe 390, MVS/ESA in the 80's and early 90's. After watching your videos about Hercules, TK4, and TK5 I am getting back to mainframe as hobbies...I enjoy your videos very much...and it is nostalgia for those times...
The peculiar requirements of ACP/TPF/ z/TPF have over the years caused IBM to develop a number of processing features to increase performance. It is always amusing to those of us that worked in this environment to observe how little it is known outside our small community. Excellent video though, some fascinating details and always good to get an overview of the developments.
Moshix, I love these videos! Not much mainframe experiece here but a lot of IBM midrange systems (S/36 and AS400). I’ve forgotten much of the minutiae after 30 years and I think every one of your videos brings back something. lol. Some of the s/36 One of my fondest memories is working at IBM Boulder as a contractor (I may still be at IBM today if they didn’t have a hiring freeze at the time). I was the “owner” of nine as/400 systems for one customer of IBM’s Global Services/Managed Operations group. My systems were located on raised floor space in building 24 and one had eleven tall cabinets. I ended up leaving and working for an IBM business partner and did many, many as/400 conversions from cisc to risc in the late 90s. Now I have Hercules running at home, on an Armbian based Quadra thanks to you!
Awesome! A great recap for us, having worked with these beauties since the start. We had (also) the PS44 one of the special operating systems, this one a scaled down PCP system aimed at the System 360 model 44 scientific mainframe. Ours had the storage ops feature, so it ran PCP fine, too.
Fascinating. I have been employed in programming since 1984. I have worked for all kind of companies in a few different countries. I have worked on all kind of systems from embedded to desktop application to cloud service. I have met other programmers socially all over the place. Yet never have I met even one programmer of IBM mainframes. They must live in some different parallel universe. Where do they hang out?
The High priest-hood of the IBM mainframe world was definitely of great bemusement to an ex-DEC/GEAC and latterly Unix-environment application developer/installer like myself. The one IBM "product" I found interesting and wowed by was APL
For about 18 months in the late 90s I was a swing shift operator on the mainframe at DHI Computing Service (Provo Utah) and if I remember right it ran VSE/ESA. Great job, very interesting. And this is a great youtube channel! :)
In 1978 I worked for IBM and used VM/370 to host a Dos/VS & CICS/DL1 server we were developing an application on. I remember sitting at two terminals, one connected to the local VM system and one connected to a system in Holland and I sent a message from one terminal to another - not that special you might think except I forced it to be sent via Australia - in took between 1 & 2 secs to go around the world, and I knew at that point that a global network for everyone was only a matter of time (turned out to be about 15 years before the internet happened). The local VM box had 16MB of memory and had 10x 100MB 3330 disks, and supported c. 50 CMS users and my DOS/VS image. For comparison, today my (Linux based) Android phone supports one user part time and has 256x as much memory and 256x as much (solid state) disk space.
Although I used numerous different systems, I did spend a lot of time on VM (CMS). One of my major accomplishments was a massive interactive graphics application via GDDM and a 3270PC AT/GX graphics workstation (we had 11 of those). I created quite extensive object/symbol libraries (mostly electronics circuits/avionics related). I also used the GDDM binary data editor functions to create new interactive editing functions so that you could actually modify drawn objects interactively (GDDM came primarily as output only but with some "transformation" editing built in). Without my editing functions, you could not change some things like line color. I had to read the binary data in the GDF segment and modify and restore it (I forget the process, it might have involved deleting the original and replacing it with a new segment, requiring me to also edit its priority to get the redraw order correct). It could also function on a 3270 terminal with programmed symbol sets used for graphics, but the interactive cursor control was done using the arrow keys and was not very satisfactory. It did work though. I had it at least as full featured as a typical mentor graphics workstation of the day (but with 16 colors instead of B&W). Some of the features I added were modeled after mentor graphics features. I still have the code (except for the object libraries). If I had a mainframe and a graphics terminal, I could resurrect it. :)
:) oh, that would be fun. Unfortunately, I'm in the middle of rebuilding a 1952 Allis Chalmers CA tractor at the moment. I expect that to take most of the year.
I have worked on every ibm os since 1974 … and on Amdahl’s and ibm mainframes … I also supported and developed a hypervisor that prior to prism and LPARS, allowed vm and MVS to share the machine in native mode each. That is MVS did not run as a guest os of vm. Vm/pe was its name… I also managed dr Amdahl’s last mainframe development at Andor Systems…
I remember TOS. There was a small shop way out in the hinterlands of Texas that was owned by the US Army but run by a contractor. I visited them in about 1985/6. They were still running a 360 with TOS. The Army commander in charge did not believe in computers. They somehow were able to attach disk storage, but had to manually map out on paper what data each track contained. Payroll on these tracks, ledger on those tracks, and so on. Those maps were a critical part of their installation documentation. About that time, the commander was convinced of the need to replace the then 20 year old machine. They brought in a 43xx along with VM. I forget which guest they installed, probably VSE. That shop was so tightly run that they converted from TOS over one weekend with not a single glitch. The 360 went back to IBM for spare parts.
Thanks for sharing! Where in Texas? I am from Texas. And yes, it was possible to write to dasd without having official support, but you had to know where to store and retrieve the data. Thanks !
I'm working at a VSE shop - 21st Century Software hired a LOT of VSE talent to get the full build process converted, and its definitely reduced the worry we had about support.
Thanks Moshix! In my first job in 1985, I programmed in Fortran in an IBM OS/environment called VSPC (Virtual Storage Personal Computing) which used a 3270 frontend and allowed us to submit batch jobs. My recollection was it was mainly just an editor and a batch submission interface. We also did our documentation using DCF with it. It was much simpler to use than ISPF/PDF on TSO but not nearly so powerful or interesting. At the time, there was also a lot of work being done on the MVS/SP machine too so I don't know if VSPC ran in an LPAR. I suppose you would classify it as a Time Sharing/Development/Testing operating system. I much preferred using VM/CMS after VSPC. But the best of course was ISPF/PDF on TSO/MVS.
The development experience was somewhat smoother In VM/CMS, XEDIT was very nice. But MVS was more interesting because with the ISPF frontend you have the full gamut of all the subsystems to look at and all the DASD datasets via the catalogue which you could see entries for even if you didn't have permission to look at the contents. I learnt a lot about MVS by rummaging around the sub systems (at a Fortune 50 company) looking at JCL procs (which kind of anchors MVS) and the tasks running in JES2 via SDSF. Those were the days 🙂
Just what I needed. I have heard of the power of the newer machines. Even installed hurcules, but never did anything with it. Amazing to here where it all came from.
I started as systems engineer at a shop running a 370/168 and one of the first 3033 under MVS/SE until at last a huge Sysplex of a couple of ES/3090 under OS/390. As at that time I could say hi to nearly every second bit in the OS, I switched to networking because it was time to clean up the mess there with at least 50 different incompatible protocols. As we said in the days: System engineers don't die, they just expire.
There was a third-party capability-based operating system for the 370 called KeyKOS. It has always stuck out to me as _the_ way to build secure and reliable systems on commodity hardware, and seems shocking that they were doing so back in the 70s. Norm Hardy made the spec available long ago and there have been attempts to replicate it.
You might have snuck in that the early 360's only had base register and 12 bit offset addressing, which was a huge pain. Any array over 4096 bytes had to be addressed in a clunky way, either by the assembly language programmer or by a compiler. That was partly forced in order to make the instructions smaller so they could feasibly sell the smaller memory models. The VM systems were tantalysing as they were a completely different and considerably more flexible design than the original OS. Tantalysing as though you could run multiple processes, they could only communicate through clunky "virtual card readers" and virtual printers. Quite elegant in some ways but still mind-numbingly restricted.
Now this is a good video! Makes me want to make a video on TSS... I don't know if anyone actually used TSS much. I got it working under VM/HPO 4.2 and 5.0, though... it seems relatively reliable but also lacking in programming languages.
I cut my teeth on TSS/360 in the mid 1970s at Carnegie-Mellon on a 360/67. I was part of the team responsible for maintaining the OS and mainframe there. Later went on to work on JES2, MVS and other products at IBM. TSS was way ahead of its time. LMK if you need any input for your video.
Sorry mate - you've missed a couple of steps at the beginning. Firstly Tapes were the big thing, and the 1st 360s ran loading separate programs from a tape drive. The tapes had a sort of Bios preloader called Basic programming Support or "BPS" for short. I only remember Fortran being supported. The compiler loaded from tape with the BPS module loading first, then the compiler, but the Fortran source was on read in on Cards. The Compiler output a copy of the BPS module onto the object tape, followed by the compiled Fortan program. Then that Tape was readied, and dialed and loaded to run the application. So 360s 1st ran under BPS. Then came Basic Operating System (BOS), which was like BPS, but only loaded once, and it executed applications as directed by cards without pausing between jobs. BOS could be IPL'd from a tape or disk, but as the use of these two BOS versions became different, IBM released BOS "Tape resident", and BOS "Disk resident". As the Disk resident version became the "goto" IBM dropped the Tape version, and renamed BOS to Disk Operating System (DOS) which was rewritten and released as DOS 2. (as I remember)
IBM DOS was the most popular OS for 360/370's mostly used on 360-30, 360-40. 360-50, 360-65 and above used OS/mft/mvt IBM would announce a machine in 1965 0r 1966 and take orders(each machine was custom built to order), then deliver these machines in 1966 0r 1967 0r 1968. WE(AS A COMPUTER BROKER) would buy and sell order positions(in 1970's). We would order 100 360's sell 50 order positions then complete the purchase of 50 360's on leases. ONE MYTH WAS THAT MOST 360'S WERE ON MONLY RENTAL a CONTRACTS, THAT WAS FALSE!!!!! ALMOST ALL WERE ON LEASE PURCHASE AGREEMENTS(FINANCED BY BANKS OR LEASE COMPANIES). When leases were up, we would expect to make a 10% profit when the customer would purchase the machines. 360's had $10,00 to $20,000 worth of gold plated contacts, so if the customer returned a 360, we sold it to tiwan gold dealers. 370 and later machines used palladium which had very little value. So a guy out in lake ronkonkama long island wound up with the problem of disposing of 600 370's in a massive whare house.(we were going to buy 50 370's and try to export them, the deal never came off, he changed the price every day!!!) lARRY was a fool, he gave free storage to leasing companies at this giant warehouse in order to get 360/370 repair and refurbish work. When the 370 market crashed he was left with the problem of having to dispose of 600 370'''s(each the size of a one bedroom apartment!!! I heard that larry went out of business, he probably had turn his warehouse into a landfill site.
Thank you very much for this excurse to the Mainframe world of OSes. I had no idea how simple that machines were in the beginning. I come from the other line of IBM machines - that midrange things called S/36, S/38 and AS/400. Interesting is, that these machines are backwards compatible until today with iOS in the System i. Memory in this machines (S/38 and AS/400) is also called "storage". Single-level-storage. But this would go too far and interests no one in the mainframe business. By the way: The System i is also capable to drive thousands of terminals and printers......
I worked on VM from about 1980-1985, first for a big insurance company (ironically on Amdahl mainframes) then for IBM. Internally, IBM upper management tolerated VM … as the story went, they prohibited any new VM/370 releases, so they were done as, “Service Packs” (SP) ostensibly fixing bugs (lack of device support was a bug, right?). As the video noted, lots of people called HPO the high priced option; it was how IBM got money if you were on a bigger mainframe - i think there was also more software optimizations. When I went to work for IBM in the early 1980s, I worked on VM/XA MA. By that time, IBM was in its’ glory; massive overhead, tons of management, just unbelievable politics. They’d finally embraced VM (we joked that IBM embracing something was how they killed things), so it was mainstream DSD/IBM; in sort, an awful place to work. I escaped DSD & worked on stuff that IBM didn’t much like (so, better work environment) until I quit in the mid 90s. BTW we also called MVS, “man vs system” since it was so, “IBM like” (overbearing). As part of VM development I actually single stepped an IBM mainframe (probably a 3081 of some variety), since I knew how to use CP - the mainframe console commands were similar (ADSTOP as I recall for setting a breakpoint).
Thanks for sharing. Yes, in those years IBM was unbearable. They acted like a government towards their customers. Nowadays IBM is irrelevant for the most part.
DPPX/370 is the DPPX/SP operating system ported to the IBM ES/9370 mainframe family of processors. DPPX/SP was a centrally managed, distributed processing system designed to run on the IBM 8100 family of processors. DPPX/SP operating system was written in a high-level language PL/DS with almost no assembler code. Once the PL/DS compiler was changed to generate mainframe hardware instructions, it made it possible to port DPPX/SP from 8100 to the mainframe. Interesting note - DPPX/370 may be the only mainframe operating system that does not make 0 to 4k addressable for both read and write to memory from application code.
I don't believe that the PL/DS compiler was changed to generate 370 instructions. PL/S (earlier it was called BSL, and later it became PL/AS) existed before the 8100, so migrating DPPX to 370 should have been easy from the compilation issues. PL/DS was a derivative of the PL/S compiler. There was a similar derivative for System/38, as I recall.
@@PeterGCapek I was on the task force to determine which compiler to use for the port of DPPX from 8100 to the mainframe. It was a shoot-out between PL/DS and PS/AS. At the time PL/DS had superior optimization compared to PL/AS. We chose PL/DS compiler over PS/AS to port DPPX to the mainframe.
Yes, IBM made a killing from under-specifying. Went for a progress meeting at a client who were moving from ICL to IBM, under DOS/VS. Having checked that we were meeting the requirements on the ICL kit, we asked how things were going IBM-wise. "This weekend we are having the Group 2 upgrade, which basically doubles the speed (and price) of what we were sold." So that will fit your needs then? "No. We'll need to double again." How was the course on DOS/VS? "It's just like UDAS on ICL, but you have to pay extra for it."
Moshix, in MVS 3.8 we have a tool called IPCS (the interactive problem control system) which apparently allows you to record and manage problems. I have been trying to get this up and running with no success - basically failing at setting up the unusual VSAM disk partition it requires. Have you any idea how this works or have any resources? I have the IBM manual for IPCS but I can't get the right VSAM files setup. I'm hoping to get this up and running to manage issues on my MVS 3.8j mainframe which I plan to make public at some point. Having system software to manage issues rather than a text document would be more useful.
According to Wikipedia it is. I own a copy. For decades now. Ought to read it some day. That, and the book I bought with it - The Limits Of Software by R. Britcher. I was told by someone on the spectacularly fail AAS project (that IBM was developing for ATC modernization at the FAA) that this other book was about that and it was useful to be familiar with the MMM book.
Hi moshix, i developed a jcl for compiling/linking cobol programs in mvs (using the standard procedures) and one for executing them, rather than using those provided with tk5 that are "compile-link-go" style (i want to have my object code in a library and use it when needed, rather than compiling every time i sub the jcl). if you are interested for publish them in a future video, please tell me. i know that a mvs developer can blindfolded write them, but for newbies they can be interesting.
Yes - it was called MDF (Multiple Domain Facility). We used it on an Amdahl 5890, in the 1980s and early 1990s, to run VM, and MVS. We also ran test MVS systems under VM.
@@moshixmainframechannel Checking my memory against wiki it seems there kind of were, but they were things you covered, with the 360/67 used 32-bit virtual addressing but a smaller physical address-space.
Only a few minutes into this, but I note that you showed CP-40 on one image, and on your early history image you started with VM/370. Having used it back then, I was under the impression that CP-40/CMS (Control Program / Cambridge Monitor System), eventually morphed into CP-67/CMS, and then CP-370/CMS, which was the predecessor to VM/370. Maybe you've mentioned this later in the video, but I was disappointed this history wasn't in your early slides. We'll see, and I may update/reply. OK, I see you do mention that on the full lineage slide, so I'm now happy. Thanks.
Used DOS/360 on a 360/30 which we upgraded to a 360/40. At the time at the insurance company across the road, the programmers had full use of a 370/158. How the other half live - their building also had a company bar!
Our (bureau) company made a fortune running the hell out of the cheapest kit to be found. We would look for a company wanting to change brands, and run their existing workload, either taking on their kit or merging onto an existing box. "Your people can be all motivated by working on the new stuff !" Then "You said the project would be completed by the end of this year. We had already lined other things up for our people. Another THREE years?? " Kerching!
ruclips.net/video/ip0cQiQjd18/видео.htmlsi=tH1tQyVXJr4Hi96y&t=1718 SAA was Systems Application Architecture and the UI part of that standard was CUA Common User Access. The idea was to unify ways of creating applications across MVS/ESA, VM/CMS, OS/2 etc. It was a bit of a meta standard that people tended to observe.
This is the lineage from the first IBM System 360. The 1401 predated the 360 and had a completely different architecture that did not use 32 bit words and 8 bit bytes as we know them today
The tittle of the video is mainframe operating systems from 1965-today. 1401 came 6-7 years before, so it’s not covered here. Also the 1401 did not have an operating system
In the early 80's I worked for a small loan finance company in Des Moines, IA. They did processing and printed statements for a large number of similar companies world wide. We would IPL a 370-145 every morning with ACP9/TPF to support some 30,000 teletype terminals processing loan payments out in their offices. (same machine would be IPL'd again after the business day to MVS to support record keeping and reporting. The shop also had a Tandem machine, Burroughs, and HP small mainframes. A 3031 supported the programmers, statement printing, record archival, etc.. It was fun to be a computer Operator back then, you felt like the center of the universe.
Thanks for sharing. Yes that was a very common setup !
Back then it wasn't called cloud computing but dialing into the mainframe or the service bureau.
I worked for a Fortune 100 food processor 1975/6 and the mainframe would crash a couple times each day. They kept score on a board on the wall and staff would plan their coffee and lunch breaks around the machine crashing.
@@KameraShy thanks. Yes it was common in those days. And it didn’t matter as much back then because computing was mostly batch and so they could easily wait
yes, I started out as an Op, it was rather high-stress at times, TBH, but a useful grounding in some aspects of the IT "biz"
Thank you for this video. It is the most comprehensive review I have ever seen. I was glad to see you included DOS/360 as when I told people I worked on DOS they assume I meant MS/DOS. They never heard of DOS/360. It was the right operating system for many users.
Absolutely it was. Did the job and wasn’t overly complex.
Dad worked for IBM. In the late 80s he had a dial in and I used VM a lot. I also wrote a REXX script to solve a problem for him and it ran in MVS.
Having used both, 15 year old me certainly preferred VM to MVS :)
A few things I remember from my time with big blue...
The Mass Storage Subsystem which was a disk equivalent of virtual memory emulating huge amount of disks using a robotic tape system. A baseball mitten was apparently an integral part of the robotics.
I once had to fix a mainframe that the customer had a major complaint about - the System Engineer had configured and ordered it with the wrong colour panels. In the end I had to order the correct paint and send the panels to a car paint shop.
Thanks for the walk down memory lane.
I started my career fresh out of high school on a RadioShack TRS-80 in 1977, and within a few months landed a job as an application programmer on a IBM 370/138 running DOS Rel 34 with CICS. It was an assembler only shop, but I learned a lot. I then took a job as a systems programmer for a DOS/VSE shop running a 4341. When DOS began to run out of gas, we upgraded to a 4381 and ran VM with multiple DOS guests. This set me up as a VM/DOS/VSE systems programmer where I changed jobs a couple of times. After several years of that work our shop upgraded to MVS, so we maintained multiple virtual DOS/VSE machines as well as standing up an MVS instance, while we migrated data and applications over to MVS. I stayed as an MVS systems programmer for about 15 years (until mid -80s), and have spent the years until now as a software engineer on Linux, Windows and occasionally some mainframe jobs. I’m still working as a software engineer, and focused on AI, Robotic process automation and system integration work.
What an amazing field to work in and I feel very privileged to have had the opportunity to spend my life doing what I love.
Good for you !! Thanks for sharing !
Fun fact: Interactive Systems Corporation had a version of Unix for the IBM 360 named ix/360. Kodak (yes, the film company) bought it in 1988 and sold it to Sun in 1991.
Wow ! Yes I heard about this 20 years ago and forgot. Thanks for reminding me !
This was very interesting and brought back a lot of memories. I worked for IBM during most of the referenced time (1970 to 2000). Started out in final test for 360/40 in Poughkeepsie. I was in the first class for training test techs on the 370/155 DAT box (version 13). Moved on to Field Engineering Division where I made a lot of money doing DAT box conversions (adding Dynamic Address Translation) to standard 370/155 aka 3155) over a weekend. In the mid 70's I went the software route servicing DOS/VS and it's successors. Yes, there were a lot of patches (PTFs - Program Temporary Fixes). Ended my career in late 1999 working in the VSE System Center (Endicott) building VSE systems (SIE - SoftwareXcel Installation Express) for VSE users. We had our own data center with a 9370 processor running VM/ESA on which we built and tested each customers VSE/ESA system (including POWER, CICS, ICCF, VTAM). Our group also hosted user group meeting around the country for VSE users where I spoke on the need to prepare for Y2K.
Oh nice! Interesting career ! So you must know a lot about the original dat box! Thanks for sharing !
1980s I worked for ITEL/NAS installing "Add on Memory Frames" to 370/155s. Hundreds of wirewraps to get another half megebyte or so of mainstorage. Also over the weekend. It paid my mortgage I guess.
I once worked on a project to keep a data center limping along until it could be completely decommissioned. Mostly changing tapes and operating printers. I got to know the guy who ran the place in its heyday. It was a large bank. They ran all their CICS and ATM transactions, as well as payroll, trust funds, etc. etc. in 512K of RAM in the 70s. When they upgraded to 1M of RAM, it was days of heaven! They had the first shared ATM network around town in the entire country. All in the RAM of an XT! My admiration for IBM is boundless.
Thanks for sharing !!
I worked on banking systems running on IBM mainframes from 1978 to 2015. Wouldn’t it be cool to expand the timeline to also include hardware names and models - processors, dasd, tape, card, printer, etc. well done Moshix, I learned a lot.
Thanks. I was thinking of adding processors in a next video
When I worked for big blue in the 80s I once was invited to go sailing by the Data Centre manager of my customer and the assistant data centre manager had a S/370 console mounted on his bathroom wall. I asked him why he hasn't made the lights flash and the reason was apparently that each bulb needed 2A to light up and there were literally hundreds of flashing lights.
Crazy waste of energy. I have an S/370 panel in my office
Thanks for the memories. I started my career in IBM systems programming back in the tail end of the OS/360 days on a 360-40. That didn't last long as they upgraded to a 4331. I moved to 43xx-30xx families and specialized in VM and MVX. I did a ton of work as a consultant to EI DuPont in Delaware at the Experiment Station (interfacing a Cray X-MP with a 3081 used as a front-end IO processor), and the textile fibers, P&EP and Berg Electronics divisions. I also dabbled in DPCX on the 8100 family. Good times.
Good times indeed
This video brought back a lot of good memories! I worked with MVS/SP to Z/OS as an operator from 1983 to ~2010. All in all it was a decent career. I still have occasional dreams of working with mainframes!
Enjoy the retirement. Thanks for watching !
Programmed a 370/115 running DOS/VSE in Assembly, COBOL, FORTRAN, RPGII, and JCL at the beginning of my computer career. I even knew how to program 029 and 129 card punches.
Nice ! Thanks for sharing!
FINALLY !! Thank you for this Video !! I waited for 242 Episodes for THIS episode. This should be Video M001 !! Next Video - The Basic Architecture of a IBM 360/370 Mainframe - CPU - IO CONTROLLER - BUS Systems - Terminal-, Disc- and Tape- Controller etc. etc. PLEASE !!! And maybe a hind where to find the right document to approach the basic understanding of a IBM 360/370 Mainframe. And No ... i know there a over a 1000 IBM documents on Bitsavers and the Internet-Archive ... i need a hind ... a roadmap what to read in what order.
I worked on DOS and DOS/VS systems as a sysprog from 1970 to 78. I installed CICS in 77 also VTAM and NCP. You know you are on the leading edge when IBM SE's ask you how you did it!! Happy Days
CICS on VSE in 77! That’s avant garde !
@@moshixmainframechannel The first CICS system I worked on was in 73! It was an "Entry" version add was full of bugs!
@martincox4520 wow. I worked with CICS 1.3 a bit. 1.6.1 was the one that truly worked well.
Fascinating, thank you for sharing that.
I'm a child of '62, started learning about (taking an interest in) computers in the late '70s. Always had this impression a Mainframe was so much more powerful than that, so much faster...
To the youngsters today, the trick to understanding a Mainframe, is the airline booking system, the hotel booking system, the credit card system.... banks, earlier versions of the stock market systems...
It is a system that has more resources than most home user / hobbyist computers, it has availability that's great. They had and worked with virtualisation since before personal computers existed. I can remember being told in about '82 that they could defragment memory, the same way you defragment a HDD, and they had two separate memory banks, and you could defrag active memory into one and de-power the other, then power it up again, then power cycle the 1st bank. If a program has a memory leak, loses track of the memory, there's only one way to recover that memory, which is to power cycle it. So for mainframes, they invented this system of having 2 separate memory banks.... It was a lot quicker to do that than it was to power-cycle the whole machine. The thing with a mainframe, was it had (let's say) 8 CPU cores, but you could shut down one or two or four, and hot plug them. Same with memory, HDDs, network interfaces, power supplies.... In a single cabinet, you had a back-plane / spine, and as long as you kept one motherboard plugged into the backplane, and power on, the machine was technically still up and could still respond to flight bookings and hotel reservations and things. The goal was that you may well have some issue with one part of the system, but there was almost nothing short of a nuclear war, that could take the whole thing down.
in 90' I work with PL/1 I found one word misspelled in the manual and IBM send me certificate thanking for the correction, I could read the manual write complicated code and run it the way I have intended too; It never happened again later in any other hardware, software systems, that is why we have Google and Stock Overflow, (PS if you had access to the red books you were a God at that time, another famous "color" book was: Brown Blue book ;) jobs ref )
Just to comment on how unique IBM's mainframe offering was, I once reviewed a UK bank's attempt to use Windows to run even a part of their processing. Just getting connectivity from 10,000 plus user PCs to a single Windows server was a major headache and required them to write a lot of special code to achieve it - IBM's mainframe was definitely the most scalable technology of its time.
In early 70s we had access to VM/370 in high school. I liked virtualization. Someone told me that someone managed to boot VM/370 inside itself 9 levels deep! Then when X86 hardware virtualization came out, I had to make it boot inside itself 10 levels and run Linux inside that (s.l.o.w.l.y) :D It was fun knitting pagetables.
I think VM/370 could only go max 2 levels deep on machines that supported VM/370. Never heard of 9 deep for VM/370.
@@moshixmainframechannel I sort of remember it was at somewhere like carnegie mellon (50 years ago), may have been chopped-up to make it work. but whoever told me (I don't remember) could have been wrong/making it up. you have to present a virtual machine with all the features that the hypervisor uses to run virtual machines.
Thanks for the history. I worked with mainframe 390, MVS/ESA in the 80's and early 90's. After watching your videos about Hercules, TK4, and TK5 I am getting back to mainframe as hobbies...I enjoy your videos very much...and it is nostalgia for those times...
Glad you are coming back to mainframes !
The peculiar requirements of ACP/TPF/ z/TPF have over the years caused IBM to develop a number of processing features to increase performance. It is always amusing to those of us that worked in this environment to observe how little it is known outside our small community. Excellent video though, some fascinating details and always good to get an overview of the developments.
Thanks! Yes, since we don’t have ACP or TPF to play with, very little is known about it. Almost nothing
Moshix, I love these videos! Not much mainframe experiece here but a lot of IBM midrange systems (S/36 and AS400). I’ve forgotten much of the minutiae after 30 years and I think every one of your videos brings back something. lol. Some of the s/36 One of my fondest memories is working at IBM Boulder as a contractor (I may still be at IBM today if they didn’t have a hiring freeze at the time). I was the “owner” of nine as/400 systems for one customer of IBM’s Global Services/Managed Operations group. My systems were located on raised floor space in building 24 and one had eleven tall cabinets. I ended up leaving and working for an IBM business partner and did many, many as/400 conversions from cisc to risc in the late 90s.
Now I have Hercules running at home, on an Armbian based Quadra thanks to you!
Thanks for sharing! Happy you are using Hercules !
Awesome! A great recap for us, having worked with these beauties since the start. We had (also) the PS44 one of the special operating systems, this one a scaled down PCP system aimed at the System 360 model 44 scientific mainframe. Ours had the storage ops feature, so it ran PCP fine, too.
Yes. Heard of PS44
Fascinating. I have been employed in programming since 1984. I have worked for all kind of companies in a few different countries. I have worked on all kind of systems from embedded to desktop application to cloud service. I have met other programmers socially all over the place. Yet never have I met even one programmer of IBM mainframes. They must live in some different parallel universe. Where do they hang out?
Here! In this channel. And in our discord !
@@moshixmainframechannel Ha! Great. There is no way I'm getting into mainframes but I'd love to follow along. Will be keeping an eye out.
The High priest-hood of the IBM mainframe world was definitely of great bemusement to an ex-DEC/GEAC and latterly Unix-environment application developer/installer like myself. The one IBM "product" I found interesting and wowed by was APL
For about 18 months in the late 90s I was a swing shift operator on the mainframe at DHI Computing Service (Provo Utah) and if I remember right it ran VSE/ESA. Great job, very interesting. And this is a great youtube channel! :)
Thank you !
In 1978 I worked for IBM and used VM/370 to host a Dos/VS & CICS/DL1 server we were developing an application on. I remember sitting at two terminals, one connected to the local VM system and one connected to a system in Holland and I sent a message from one terminal to another - not that special you might think except I forced it to be sent via Australia - in took between 1 & 2 secs to go around the world, and I knew at that point that a global network for everyone was only a matter of time (turned out to be about 15 years before the internet happened).
The local VM box had 16MB of memory and had 10x 100MB 3330 disks, and supported c. 50 CMS users and my DOS/VS image. For comparison, today my (Linux based) Android phone supports one user part time and has 256x as much memory and 256x as much (solid state) disk space.
Exactly. We do the same stuff as back then but it requires 256x as many resources because we point and click now
Although I used numerous different systems, I did spend a lot of time on VM (CMS). One of my major accomplishments was a massive interactive graphics application via GDDM and a 3270PC AT/GX graphics workstation (we had 11 of those). I created quite extensive object/symbol libraries (mostly electronics circuits/avionics related). I also used the GDDM binary data editor functions to create new interactive editing functions so that you could actually modify drawn objects interactively (GDDM came primarily as output only but with some "transformation" editing built in). Without my editing functions, you could not change some things like line color. I had to read the binary data in the GDF segment and modify and restore it (I forget the process, it might have involved deleting the original and replacing it with a new segment, requiring me to also edit its priority to get the redraw order correct). It could also function on a 3270 terminal with programmed symbol sets used for graphics, but the interactive cursor control was done using the arrow keys and was not very satisfactory. It did work though. I had it at least as full featured as a typical mentor graphics workstation of the day (but with 16 colors instead of B&W). Some of the features I added were modeled after mentor graphics features. I still have the code (except for the object libraries). If I had a mainframe and a graphics terminal, I could resurrect it. :)
Well, we have GDDM capable terminals and we have VM with GDDM. Contact me on discord and I will set you up so you can revive your software
:) oh, that would be fun. Unfortunately, I'm in the middle of rebuilding a 1952 Allis Chalmers CA tractor at the moment. I expect that to take most of the year.
I have worked on every ibm os since 1974 … and on Amdahl’s and ibm mainframes … I also supported and developed a hypervisor that prior to prism and LPARS, allowed vm and MVS to share the machine in native mode each. That is MVS did not run as a guest os of vm. Vm/pe was its name… I also managed dr Amdahl’s last mainframe development at Andor Systems…
Interesting. Where can I find more info on VM/PE ?
I remember TOS. There was a small shop way out in the hinterlands of Texas that was owned by the US Army but run by a contractor. I visited them in about 1985/6. They were still running a 360 with TOS. The Army commander in charge did not believe in computers. They somehow were able to attach disk storage, but had to manually map out on paper what data each track contained. Payroll on these tracks, ledger on those tracks, and so on. Those maps were a critical part of their installation documentation. About that time, the commander was convinced of the need to replace the then 20 year old machine. They brought in a 43xx along with VM. I forget which guest they installed, probably VSE. That shop was so tightly run that they converted from TOS over one weekend with not a single glitch. The 360 went back to IBM for spare parts.
Thanks for sharing! Where in Texas? I am from Texas. And yes, it was possible to write to dasd without having official support, but you had to know where to store and retrieve the data. Thanks !
Thanks, learned a lot. Quite an amazing history, and lots of changes and innovation over the years.
Thanks for watching ! Come back soon
I'm working at a VSE shop - 21st Century Software hired a LOT of VSE talent to get the full build process converted, and its definitely reduced the worry we had about support.
Nice
Thanks Moshix! In my first job in 1985, I programmed in Fortran in an IBM OS/environment called VSPC (Virtual Storage Personal Computing) which used a 3270 frontend and allowed us to submit batch jobs. My recollection was it was mainly just an editor and a batch submission interface. We also did our documentation using DCF with it. It was much simpler to use than ISPF/PDF on TSO but not nearly so powerful or interesting. At the time, there was also a lot of work being done on the MVS/SP machine too so I don't know if VSPC ran in an LPAR. I suppose you would classify it as a Time Sharing/Development/Testing operating system. I much preferred using VM/CMS after VSPC. But the best of course was ISPF/PDF on TSO/MVS.
Agreed. VM/CMS is a lot of fun. Thanks for writing !
The development experience was somewhat smoother In VM/CMS, XEDIT was very nice. But MVS was more interesting because with the ISPF frontend you have the full gamut of all the subsystems to look at and all the DASD datasets via the catalogue which you could see entries for even if you didn't have permission to look at the contents. I learnt a lot about MVS by rummaging around the sub systems (at a Fortune 50 company) looking at JCL procs (which kind of anchors MVS) and the tasks running in JES2 via SDSF. Those were the days 🙂
@drigans2065 yes. That’s how we all learned. Looking arround until somebody told us to stop
Worked at Data center on IBM 3030 later 3032 systems with OS/MVS SPF & TSO.
Punch cards on Fortran 77 and lots of JCL.
3032 were fine machines with good support for MVS/SP. Thanks for sharing
Just what I needed. I have heard of the power of the newer machines. Even installed hurcules, but never did anything with it. Amazing to here where it all came from.
Thanks 🙏
In VSE/AF (or DOS/VSE/AF as it is referred to in our mainframe history document by Dave Morton) AF stands for Advanced Functions.
I started as systems engineer at a shop running a 370/168 and one of the first 3033 under MVS/SE until at last a huge Sysplex of a couple of ES/3090 under OS/390. As at that time I could say hi to nearly every second bit in the OS, I switched to networking because it was time to clean up the mess there with at least 50 different incompatible protocols. As we said in the days: System engineers don't die, they just expire.
Nice! Thanks for sharing !
There was a third-party capability-based operating system for the 370 called KeyKOS. It has always stuck out to me as _the_ way to build secure and reliable systems on commodity hardware, and seems shocking that they were doing so back in the 70s. Norm Hardy made the spec available long ago and there have been attempts to replicate it.
Been working as cobol, cics, db2 mainframe programmer since 1993.
You might have snuck in that the early 360's only had base register and 12 bit offset addressing, which was a huge pain. Any array over 4096 bytes had to be addressed in a clunky way, either by the assembly language programmer or by a compiler. That was partly forced in order to make the instructions smaller so they could feasibly sell the smaller memory models.
The VM systems were tantalysing as they were a completely different and considerably more flexible design than the original OS. Tantalysing as though you could run multiple processes, they could only communicate through clunky "virtual card readers" and virtual printers. Quite elegant in some ways but still mind-numbingly restricted.
Now this is a good video! Makes me want to make a video on TSS... I don't know if anyone actually used TSS much. I got it working under VM/HPO 4.2 and 5.0, though... it seems relatively reliable but also lacking in programming languages.
Thanks. I deliberately left out TSS because it was so short lived and neither fish nor meet.
But please make a video about TSS!
@@moshixmainframechannel I'm working on it right now... I'll be sure to demonstrate the COBOL and FORTRAN compilers if I can!
@hs-tc cool! Thanks. Can’t wait
I cut my teeth on TSS/360 in the mid 1970s at Carnegie-Mellon on a 360/67. I was part of the team responsible for maintaining the OS and mainframe there. Later went on to work on JES2, MVS and other products at IBM. TSS was way ahead of its time. LMK if you need any input for your video.
That's an interesting lecture if I ever saw one, well done and the quality is very good.
Oh thank you !
Sorry mate - you've missed a couple of steps at the beginning. Firstly Tapes were the big thing, and the 1st 360s ran loading separate programs from a tape drive. The tapes had a sort of Bios preloader called Basic programming Support or "BPS" for short. I only remember Fortran being supported. The compiler loaded from tape with the BPS module loading first, then the compiler, but the Fortran source was on read in on Cards. The Compiler output a copy of the BPS module onto the object tape, followed by the compiled Fortan program. Then that Tape was readied, and dialed and loaded to run the application. So 360s 1st ran under BPS. Then came Basic Operating System (BOS), which was like BPS, but only loaded once, and it executed applications as directed by cards without pausing between jobs. BOS could be IPL'd from a tape or disk, but as the use of these two BOS versions became different, IBM released BOS "Tape resident", and BOS "Disk resident". As the Disk resident version became the "goto" IBM dropped the Tape version, and renamed BOS to Disk Operating System (DOS) which was rewritten and released as DOS 2. (as I remember)
Didn’t miss. I did say there were other operating systems. But thanks for sharing
IBM DOS was the most popular OS for 360/370's mostly used on 360-30, 360-40. 360-50, 360-65 and above used OS/mft/mvt
IBM would announce a machine in 1965 0r 1966 and take orders(each machine was custom built to order), then deliver these machines
in 1966 0r 1967 0r 1968. WE(AS A COMPUTER BROKER) would buy and sell order positions(in 1970's). We would order 100 360's
sell 50 order positions then complete the purchase of 50 360's on leases. ONE MYTH WAS THAT MOST 360'S WERE ON MONLY RENTAL a
CONTRACTS, THAT WAS FALSE!!!!! ALMOST ALL WERE ON LEASE PURCHASE AGREEMENTS(FINANCED BY BANKS OR LEASE COMPANIES).
When leases were up, we would expect to make a 10% profit when the customer would purchase the machines. 360's had $10,00 to
$20,000 worth of gold plated contacts, so if the customer returned a 360, we sold it to tiwan gold dealers. 370 and later machines
used palladium which had very little value. So a guy out in lake ronkonkama long island wound up with the problem of disposing of
600 370's in a massive whare house.(we were going to buy 50 370's and try to export them, the deal never came off, he changed the price
every day!!!) lARRY was a fool, he gave free storage to leasing companies at this giant warehouse in order to get 360/370 repair
and refurbish work. When the 370 market crashed he was left with the problem of having to dispose of 600 370'''s(each the size of
a one bedroom apartment!!! I heard that larry went out of business, he probably had turn his warehouse into a landfill site.
@edmartin6245 once VM/370 became available that became the most popular OS on IBM mainframes for a long time.
Thank you very much for this excurse to the Mainframe world of OSes.
I had no idea how simple that machines were in the beginning.
I come from the other line of IBM machines - that midrange things called S/36, S/38 and AS/400.
Interesting is, that these machines are backwards compatible until today with iOS in the System i.
Memory in this machines (S/38 and AS/400) is also called "storage". Single-level-storage. But this would go too far and interests no one in the mainframe business.
By the way: The System i is also capable to drive thousands of terminals and printers......
You are welcome!
I worked on VM from about 1980-1985, first for a big insurance company (ironically on Amdahl mainframes) then for IBM. Internally, IBM upper management tolerated VM … as the story went, they prohibited any new VM/370 releases, so they were done as, “Service Packs” (SP) ostensibly fixing bugs (lack of device support was a bug, right?). As the video noted, lots of people called HPO the high priced option; it was how IBM got money if you were on a bigger mainframe - i think there was also more software optimizations. When I went to work for IBM in the early 1980s, I worked on VM/XA MA. By that time, IBM was in its’ glory; massive overhead, tons of management, just unbelievable politics. They’d finally embraced VM (we joked that IBM embracing something was how they killed things), so it was mainstream DSD/IBM; in sort, an awful place to work. I escaped DSD & worked on stuff that IBM didn’t much like (so, better work environment) until I quit in the mid 90s. BTW we also called MVS, “man vs system” since it was so, “IBM like” (overbearing). As part of VM development I actually single stepped an IBM mainframe (probably a 3081 of some variety), since I knew how to use CP - the mainframe console commands were similar (ADSTOP as I recall for setting a breakpoint).
Thanks for sharing. Yes, in those years IBM was unbearable. They acted like a government towards their customers. Nowadays IBM is irrelevant for the most part.
DPPX/370 is the DPPX/SP operating system ported to the IBM ES/9370 mainframe family of processors. DPPX/SP was a centrally managed, distributed processing system designed to run on the IBM 8100 family of processors. DPPX/SP operating system was written in a high-level language PL/DS with almost no assembler code. Once the PL/DS compiler was changed to generate mainframe hardware instructions, it made it possible to port DPPX/SP from 8100 to the mainframe. Interesting note - DPPX/370 may be the only mainframe operating system that does not make 0 to 4k addressable for both read and write to memory from application code.
I don't believe that the PL/DS compiler was changed to generate 370 instructions. PL/S (earlier it was called BSL, and later it became PL/AS) existed before the 8100, so migrating DPPX to 370 should have been easy from the compilation issues. PL/DS was a derivative of the PL/S compiler. There was a similar derivative for System/38, as I recall.
@PeterGCapek ok
@@PeterGCapek I was on the task force to determine which compiler to use for the port of DPPX from 8100 to the mainframe. It was a shoot-out between PL/DS and PS/AS. At the time PL/DS had superior optimization compared to PL/AS. We chose PL/DS compiler over PS/AS to port DPPX to the mainframe.
Yes, IBM made a killing from under-specifying. Went for a progress meeting at a client who were moving from ICL to IBM, under DOS/VS. Having checked that we were meeting the requirements on the ICL kit, we asked how things were going IBM-wise. "This weekend we are having the Group 2 upgrade, which basically doubles the speed (and price) of what we were sold." So that will fit your needs then? "No. We'll need to double again."
How was the course on DOS/VS? "It's just like UDAS on ICL, but you have to pay extra for it."
Moshix, in MVS 3.8 we have a tool called IPCS (the interactive problem control system) which apparently allows you to record and manage problems. I have been trying to get this up and running with no success - basically failing at setting up the unusual VSAM disk partition it requires.
Have you any idea how this works or have any resources?
I have the IBM manual for IPCS but I can't get the right VSAM files setup. I'm hoping to get this up and running to manage issues on my MVS 3.8j mainframe which I plan to make public at some point. Having system software to manage issues rather than a text document would be more useful.
Never used it on MVS 3.8 before
Excellent work as usual!
Thanks !!
Was not OS/360 what "The Mythical Man Month" was written about?
According to Wikipedia it is. I own a copy. For decades now. Ought to read it some day. That, and the book I bought with it - The Limits Of Software by R. Britcher. I was told by someone on the spectacularly fail AAS project (that IBM was developing for ATC modernization at the FAA) that this other book was about that and it was useful to be familiar with the MMM book.
@@nickpalance3622 I read it many years ago and I remember it had a big influence on my understanding of how to organise software development.
SAA Systems Application Architecture
For a second I thought this was CLABRETRO. (based on the music)
New intro music :)
Hi moshix, i developed a jcl for compiling/linking cobol programs in mvs (using the standard procedures) and one for executing them, rather than using those provided with tk5 that are "compile-link-go" style (i want to have my object code in a library and use it when needed, rather than compiling every time i sub the jcl). if you are interested for publish them in a future video, please tell me. i know that a mvs developer can blindfolded write them, but for newbies they can be interesting.
Sure send along. However there are command link and store procedures for cobol already in TK5
@@moshixmainframechannel thank you, i searched but i can't find them, i only found procedures, not ready-to-use jcl...
Yes that’s what I also wrote: procedures. But from the procedure you can make a job in seconds
Wasn't it Amdahl that introduced LPARs?
Possible
Yes - it was called MDF (Multiple Domain Facility). We used it on an Amdahl 5890, in the 1980s and early 1990s, to run VM, and MVS. We also ran test MVS systems under VM.
Weren't there also IBM 32-bit operating systems for that one or few 32-bit systems?
Not as far as I know
@@moshixmainframechannel Checking my memory against wiki it seems there kind of were, but they were things you covered, with the 360/67 used 32-bit virtual addressing but a smaller physical address-space.
There was no 32 bit addressing ever. Only 31 bit
@@moshixmainframechannel The 360/67 offered both 24-bit and 32-bit addressing.
@@brucereynolds7009 well, the pointers were only 24 bit, but the word size was 32bit. As I mention on the video
Only a few minutes into this, but I note that you showed CP-40 on one image, and on your early history image you started with VM/370. Having used it back then, I was under the impression that CP-40/CMS (Control Program / Cambridge Monitor System), eventually morphed into CP-67/CMS, and then CP-370/CMS, which was the predecessor to VM/370. Maybe you've mentioned this later in the video, but I was disappointed this history wasn't in your early slides. We'll see, and I may update/reply. OK, I see you do mention that on the full lineage slide, so I'm now happy. Thanks.
CP/40 did not morph into anything. CP/67 was a complete rewrite.
Used DOS/360 on a 360/30 which we upgraded to a 360/40. At the time at the insurance company across the road, the programmers had full use of a 370/158. How the other half live - their building also had a company bar!
Unfair !!
Our (bureau) company made a fortune running the hell out of the cheapest kit to be found. We would look for a company wanting to change brands, and run their existing workload, either taking on their kit or merging onto an existing box. "Your people can be all motivated by working on the new stuff !" Then "You said the project would be completed by the end of this year. We had already lined other things up for our people. Another THREE years?? " Kerching!
ruclips.net/video/ip0cQiQjd18/видео.htmlsi=tH1tQyVXJr4Hi96y&t=1718
SAA was Systems Application Architecture and the UI part of that standard was CUA Common User Access. The idea was to unify ways of creating applications across MVS/ESA, VM/CMS, OS/2 etc. It was a bit of a meta standard that people tended to observe.
Indeed.
Wasn’t the 1401 the first real mainframe in 59? Was some sort of Fortran code right?
I think it depends onn what you count as a mainframe, there were the 701 etc before that. The 1401 etc didn't really use an OS though
This is the lineage from the first IBM System 360. The 1401 predated the 360 and had a completely different architecture that did not use 32 bit words and 8 bit bytes as we know them today
The tittle of the video is mainframe operating systems from 1965-today. 1401 came 6-7 years before, so it’s not covered here. Also the 1401 did not have an operating system
Dark Mode/background, use it
Ok
AIX ran on mainframes? Sure?
Yes, sure. We had it running and I logged in.