Outstanding! It brings light to my 1986 Introduction to Cobol for Business Administration in the University of Costa Rica. We didn't even had a textbook. Not even a tour to the Registrar's Office where from the window I could see the disk units of the System 36 mini-computer. To us it was the next best thing to a mainframe! But at the same time, the dawning the microcomputer was to eclipse this course. We could see on screen the results of a payroll designed in a Smart SpreadSheet. Dad was developing in BASIC an automated quote for his printing shop on a Tandy TRS80 with a lot of difficuty and hardware issues. Later my brother in his cargo ship business brought an Apple Macintosh Plus to his office...All this comes to mind after reading a book dad had in his library called The American Challenge by French writer Jean-Jacques Servan-Schreiber where he described the investment that IBM did in the 1960's for its System 360 mainframes of US$5 billion of that era, more than the Space Program and the adoption of the I.C. to their systems!
I worked on S/360 & S/370 systems starting in about 1975 in systems programming. I owned an S/360 model 30, 2311s and 2401 tapes at one point. Did you know the 2311 and 2314 both had hydraulic head seek mechanisms? The 3330 "pizza oven" was the first with an electronic voice-coil head seek mechanism and needed a 3830 control unit. S/360s had selector channels and byte multiplexor channels, but block [multiplexor] channels didn't come along until S/370. The 3344 disk was left out; it was a fixed pack version of the 3340. 3380s were referred to as: single-density (no suffix, D & J); Double-density (E); Triple-density (K) The disk model nomenclature could use some work. A-models are head-of-string units, while B-models only attached to A-models. The number suffix is physical CUU addresses per box. Thus an example would be a 3380-AD4 or 3380-BK4 or 3380-CJ2. C-models had a built-in control unit.
At my times @ EMC we 've had a feature called "SRDF" for disaster recovery. We copied disk unit data from one site to antoher in real time. Typically done via fibre or what you call it "ESCON". Failover scrips were provided by EMC in coop with the customer. Even Sysplex was covered.
I never used a machine like that, but they always fascinated me. I basically started my work with web applications like many (did a little bit of industrial and an algebraic system before but they were projects, and the algebra system was free for a university). Thanks for sharing. From your table, I can see why by 87 the continuation of the line was dubious as arrays started to appear and production of 80 megabytes drives was massive. Far smaller and less reliable that the mainframe drives, but too cheap to ignore.
The 3340 had two 30 MB (approx) drives side by side, or 30-30. The lever action Winchester was a 30-30 caliber. The Winchester drive was named after the gun that won the West.
Beg to differ about the "gun that won the west". The 30-30 first came out for the Winchester model 94 (1894). The Winchester 1873 would be a better candidate, and it commonly used 44-40 ammo (44 caliber, 40 grains of black powder). Winchester called the 1873 model "the gun that won the West."
In the UNIX and LINUX world tapes (at least earlier) could be very picky with regards to real time. A streaming device works best if the host can keep up with it - let the device wait too long for data means that the tape must be stopped and reversed a bit so it was very important that the data to be written either could be read sequentially or it if it was from different parts of the disk staged into a continous large file on a fast disk device, fast enough that the tape device's real time demands could be meet.
I worked with 3370 and 3380 drives in places I worked. Funny thing about the 3370, some of those would make a sound like a buzz saw once in a while, did they ever break down, uh nope. Once saw a 3380 catch fire, it was the main system volume for VM/370, did VM/370 crash, nope, it simply said the volume was offline, when we hit the big red switch. That system continued to run until 8pm, when the system was shutdown, the backup tape loaded it on another volume, performed an IPL, and the next day the techs were on-site to repair the drive.
Never had the chance to work with mainframes but when I studied to be ready if the chance arose I was told big iron systems have so much redundancy one could claim you could physically disable or destroy half of it and it would still keep running in some capacity. Wish Windows and sometimes even Linux had that level of fault tolerance, a literal lp0 printer on fire scenario without crashing or major disruption
VM sysprog/fanboy here. For reliability we had a VM system that ran continuously for almost a year without shutdown. The only reason it didn't run longer was we had to apply some maintenance that required a restart.
Coming from Win and Linux, I always wanted to learn about mainframe computing. This channel together with tk4/5 make a great resource. Remember that commercial „my own IBM computer. Imagine that“?
The IBM 1311 and 2311 were hydraulic actuated. I remember that if they were not squarely mounted on the floor the power of the actuator moving between cylinders would walk the driver across the floor., More than once we had a hydraulic line or connection fail and the device would be flooded with hydraulic fluid. 7.25 MB of storage seems minuscule in size today but it could held our compilers and programs and meant we didn't have to load them from cards every time we wanted to run them.
when I was eleven years old, the math-science center had an IBM 1620 with an attached 2311 hard drive. It was obsolete (donated by philip morris), but made great Snoopy calendar printouts. did some fortran code for that computer. lots of blinkenlights. saw an operational IBM 1401 at the defense general supply center in chesterfield VA in 1973. was briefly an IBM/360 operator in 1979 for a financial services company.
I was a S/370 console operator for a large insurance company that had a 3032 mainframe and a vast computer room full of these 3350s. (I was only 19 years old) when we would start up batch processing in the evening, it was as if a hundred washing machines hit the spin cycle all at once. Mainframe operator was my favorite job ever. Did that from '79 through '83, then became a systems programmer, i.e., installer and sysadmin. left the mainframe world in '87 never to return.
Yes the fun we had in those days is just not available anymore. You felt that it was an important job and to be taken seriously but it was still a lot of fun
@@moshixmainframechannel I was 19 and making $24k / year plus full benefits with no college degree (1979). most of my friends earned half that amount at their jobs. everyone in DP operations in the town I lived in knew everyone else. job hopping usually meant a 15% raise. I do recall that everyone working 3rd shift everywhere I ever worked was C R A Z Y - mostly in a good way. the execs never had any idea what was going on after midnight behind locked doors in the machine room . . . .
TOTALLY ...and the tape drives right behind,, acting like banks of clothes dryers. It definitely was like working at the Bizarro Jetsons Laundromat. Tape Ape ,and proud of it...
Just retired. Started work in IT in 1976 on an IBM S360/40 Have used all these DASD systems except 3340. I remember when our site reached 1.4TB of online DASD. (3380's & 3390's)
Sounds like a Canadian guy I know…. I will retire soon after 47 years. Cheers Paul This stuff takes me down memory lane. Moshix would like the unofficial museum in POK
For quite some time in the late 1980s, IBM subcontracted manufacturing their disk arrays to Hitachi Computer in Kanagara, Japan. They were able to produce them with only a one micron head to disk clearance which used less power and had faster read write speeds. IBM then licensed the tech from Hitachi and then manufactured their own.
I don't think the timeline is right. The IBM General Products Division in San Jose, CA, manufactured the disk equipment. Their first RAID disk product was called the RAMAC (reuse of the 1950s disk name), announced in June of 1994, and was manufactured by IBM competitor Storage Technology Corp for IBM after EMC Corporations RAID disk products cut deeply into IBM's sales. IBM announced the sale of its Storage Technology Division to Hitachi in 2002.
@@moshixmainframechannel You are correct about the 305 Ramac predating System/360, but IBM reused the Ramac name, rushing a repackaged competitor's product to market. Their first RAID disk product was called the RAMAC (reuse of the 1950s 305 disk name), announced in June of 1994, and was manufactured for IBM by competitor Storage Technology Corp after EMC Corporations RAID disk products cut deeply into IBM's DASD sales. I don't remember the 1994 Ramac's machine number. I didn't like the product and was recommending EMC's cached RAID arrays to my consulting clients.
I used them in the middle 80's they were used in Buenos Aires University, to store Student records. We called them sandwicheras ( sandwich storage ), because they were pretty much like the ones used in bars.I still remember the VOL names "BAGUAL" and "MANIJA", and the slang "to scratch a disk"
Very nice video with good technical info and slides. This would also be a good video to use in a computer science course to educate someone on the history of this media. Nicely done.
History: First removal disk pack 1311(went on 1440 computers. For 360 ibm renumbered 1311 changed to 2311 disk drive.(really same drive, higher density). Winchester drives were developed at IBM lab located in winchester england. 3330 disk packs cost $1,000 per pack(100 meg pack!!!!).
I worked with IBM, as a language instructor, at the Hursley Park Labs during the time the Winchester Disk Systems were designed and developed (in the 1960's). The programming language PL/1 was also developed at that location and I played a part in writing some compiler routines for that much denigrated development tool. Those were great, exciting, days to work at IBM when the company, headed by Tom Watson Jr, was on track to become the world's biggest company. Where did it all go wrong?
@jimhardy2034 exciting experience !! Thanks for sharing. Where it went wrong? Easy: IBM always had an attitude that doesn’t work anymore nowadays: i.e. that they know better than the customer what the customer needs. That’s what’s wrong with IBM. In other words, arrogance
I'm pretty sure that drums were used as I/O devices. I remember taking a trip to the American Airlines reservations system in the '60s and gazing at their 7330 drum attached to their 7090. It's, essentially, a head-per-track disk, no seek time. Also, all the virtual memory 360s and 370s pretty much had to have 2301 or 2303 drums for paging devices if they ran CP67 or VM. Unlike HyperV (somebody correct me if I'm wrong) CP and VM can page the virtual machines they create not keep the whole virtual machine in memory.
On the 360 drums were channel attached and used as I/O devices. Some early drums used drums for main memory. The 7090 used core memory, so indeed the drum was an I/O device.
I believe that the System Managed Storage feature provides the badly needed abstractin needed to provde potability to keep up with technology. It provides huge improvement in JCL simplification. Also, with move towards database systems like DB2, Oracle and My SQL, separation from the nature of the physical device record layout is complete. Whenever the data type of an item is changed, ,added or remvoed or archived, the systems will deal without it without extra programming!
@@moshixmainframechannel I never had the opportunity to learn what items that were the opbject or storage administration setup. When I was programming for apps on the Prime supermini systems, we were never exposed to the nature of device physical records and how many of the physical records were contained on a device track!
did you mention that an HLQ had a max length of 8 characters? I may have missed it. 8.3 filenames, meet 8.8.8.8.8 filenames, LOL Lovely history. I was "born" into mainframes in 1980, we timeshared, and had two 3350 packs assigned for our use. When we in-housed 2 years later, we bought a 4381, four 3420 tape drives, two 3380s, a 3705 NCP, and a 2501 card reader (yes a card shop!). we also rented two 3350s for about a year, to help us migrate our jobs. First CLIST i think i wrote, was, drum roll, a blocksize calculator (hello, world!), and yeah, those 3350 numbers stick with you. When i'm old and demented, i'll probably still remember 555 x 10 x 19069, even though it's been 39 years! My first sysgen was a CBIPO of MVS/SP for that move - and a PEPgen for the RJE remotes and dial-out stuff.
Hi Moshix, love your channel, great fun playing with these vintage systems on my windows PC (even though I work on various z/OS systems professionally it's difficult to play with things outside my specialisms). I was wondering - whether you have considered doing a video on setting up applications that use started tasks. Is anything available to be downloaded? (I know there is KICKS, but I'm unsure whether that can be set up to run as an STC in MVS3.8 in the way CICS can under z/OS). Or, is it possible to create a simple user-defined app that runs as an STC under MVS3.8? I'm just thinking of the basics of how to define the JCL/procudures, stopping/starting tasks, maybe a bit of automation, APF authorisations etc. Anyway keep up the great work really appreciate the videos, always dense with history and knowledge to learn from.
Hey Moshix, a topic for future consideration - I'd really like to get into the basics of interactive mainframe apps on MVS 3.8 and the available languages and tools we could use today to develop interactive panels, including menus and data entry forms. I know you've covered the basics of installing KICKS in the past and I'm about to try that myself; does REXX for MVS 3.8 also have the capability to author panels, and what other tools could we consider, hack around with, study from, or learn about, for example how was the TK4- menu system developed?
Well with the new BREXX 2.3 it’s easy to build panels in TSO in MVS3.8. But you would have to wait for Update 9 of MVS 3.8 TK4. Hopefully out by the end of the year.
The good old DS6800 as photographed for the present day. However even though I have one I will be restoring I would never trust one in production. The main reason is the fact of the KONA cards fail at random without warning and if I properly quiesced they will sometimes spontaneously format. But anyway I would trust my data more to a DS8K line of hardware.
Yes which is why I’m looking into dumping it to tape and transferring it to a DS8800 that was decommissioned from the Miami university. The system was running 2 VSE LPAR in ESA/390 mode. Only one issue is I need to buy replacements disks for the DS8800 and the CECs has the disks removed but I believe the previous CE left the restore media so I’m looking into restoring it. However when I’m all said and done if anyone wants that DS6k with CKD Licence I am looking into selling it everything disks software and expansion for the price of 1 KONA.
Hi moshix, not related to this video, but I was hoping you could point me to the best IBM Assembly Language reference (in PDF form) which runs under MVS 3.8j. I found some on bitsavers but was hoping you could name "the one" you like most. Thanks!
You skipped the 2310, which is the single-platter removable disk (the platter was the 2315), developed for the 1130 and 1800 (where it was a/k/a 1810) but also used in the 360/44. DEC adopted the drive as the RK-05.
DASD, pronounced "daz dee" is direct access storage device... I think you might have said attached. Anyway fantastic video as always but I've particularly enjoyed your hardware related stories. You might want to investigate other CPU devices such as console devices and even the motor generator sets the accompanied processors from at least as early as the 135 through to the 4300 series, converting 50/60 Hz to 400 Hz. The were typically used as tables in machine rooms as they were about the same size as a 4381 CPU.
22:50 your photo is not a 3330. The 3330 was a pizza oven configuration like the 2314. The model 1 and 11 packs were different and not interchangeable.
"virtual memory wouldn't exist without fast disk" - wellllllllllll, it probably WOULD, because it was so much of a bargain to be able to swap work in-and-out-of-memory, and run "more work than your CPU could run" - but it would have been less performance, definitely. we'd be looking at hours of backlogs that we didn't have to look at with fast drives.... but you know how "money talks," and if the company didn't have to buy more capacity, they'd put up with things being slower...within reason. (p.s. at least MY company would make that compromise --- but they were cheapskates! ;)
@@moshixmainframechannel well, if no disks were available at all, then that would be gruesome indeed. I didn't think you meant, that, just that a shop might not allocate fastest disk. Fortunately it's an academic exercise we never had to try, but i always wondered about the possibility of having a tape or two permanently mounted, (just for swaps, not for paging of-course,) if you had to extend memory in a disk-free system, and could tolerate a region-swap taking a minute or maybe more. But, now I think of it, I don't believe ASM has the methods available, to manage a set of sequential-media "swap files" anyway, so scratch that idea, heh heh.
Swapping to tape did indeed exist. But those were not virtual memory address spaces. It was just memory segments. ICTS on the 7094 is an example of that
I worked at a shop in 1967 where we rolled a BG partition out to tape, loaded a transient program in from the same tape, rewound the tape while the transient program was running, then reloaded the original BG batch program back in to continue execution...........a form of virtual memory on tape. IBM did something similar but to disk IIRC
There are so many errors in here that I don't even know where to start. My credentials are 46 years as an IBM mainframe administrator/programmer from 1966 on the 360/40 through the Z series and 2311 through 3390. You can see me in my icon/avatar seated at an IBM 360/44 running 44/MFT in 1967.
@@moshixmainframechannel Agree and I don't mind corrections in the comments, I always read the comments to see the "errata". This dude just say "Many things are wrong" but don't mention a single one, and he is harsh about nothing. Thanks for sharing the info you have, even if you have an error at least you are preserving and spreading information, for free.
If I'd spent 46 years of my life wrangling IBM mainframes I'd be cranky too. I spend a couple years in VM storage management at the start of my IBM career.
Outstanding! It brings light to my 1986 Introduction to Cobol for Business Administration in the University of Costa Rica. We didn't even had a textbook. Not even a tour to the Registrar's Office where from the window I could see the disk units of the System 36 mini-computer. To us it was the next best thing to a mainframe! But at the same time, the dawning the microcomputer was to eclipse this course. We could see on screen the results of a payroll designed in a Smart SpreadSheet. Dad was developing in BASIC an automated quote for his printing shop on a Tandy TRS80 with a lot of difficuty and hardware issues. Later my brother in his cargo ship business brought an Apple Macintosh Plus to his office...All this comes to mind after reading a book dad had in his library called The American Challenge by French writer Jean-Jacques Servan-Schreiber where he described the investment that IBM did in the 1960's for its System 360 mainframes of US$5 billion of that era, more than the Space Program and the adoption of the I.C. to their systems!
I worked on S/360 & S/370 systems starting in about 1975 in systems programming. I owned an S/360 model 30, 2311s and 2401 tapes at one point.
Did you know the 2311 and 2314 both had hydraulic head seek mechanisms? The 3330 "pizza oven" was the first with an electronic voice-coil head seek mechanism and needed a 3830 control unit.
S/360s had selector channels and byte multiplexor channels, but block [multiplexor] channels didn't come along until S/370.
The 3344 disk was left out; it was a fixed pack version of the 3340.
3380s were referred to as: single-density (no suffix, D & J); Double-density (E); Triple-density (K)
The disk model nomenclature could use some work. A-models are head-of-string units, while B-models only attached to A-models. The number suffix is physical CUU addresses per box. Thus an example would be a 3380-AD4 or 3380-BK4 or 3380-CJ2. C-models had a built-in control unit.
At my times @ EMC we 've had a feature called "SRDF" for disaster recovery.
We copied disk unit data from one site to antoher in real time.
Typically done via fibre or what you call it "ESCON".
Failover scrips were provided by EMC in coop with the customer.
Even Sysplex was covered.
I never used a machine like that, but they always fascinated me. I basically started my work with web applications like many (did a little bit of industrial and an algebraic system before but they were projects, and the algebra system was free for a university). Thanks for sharing. From your table, I can see why by 87 the continuation of the line was dubious as arrays started to appear and production of 80 megabytes drives was massive. Far smaller and less reliable that the mainframe drives, but too cheap to ignore.
Excellent! Mainframe charts and mainframe history, two of my favorite things!
The 3340 had two 30 MB (approx) drives side by side, or 30-30. The lever action Winchester was a 30-30 caliber. The Winchester drive was named after the gun that won the West.
Thx
Beg to differ about the "gun that won the west". The 30-30 first came out for the Winchester model 94 (1894). The Winchester 1873 would be a better candidate, and it commonly used 44-40 ammo (44 caliber, 40 grains of black powder). Winchester called the 1873 model "the gun that won the West."
In the UNIX and LINUX world tapes (at least earlier) could be very picky with regards to real time.
A streaming device works best if the host can keep up with it - let the device wait too long for data means that the tape must be stopped and reversed a bit so it was very important that the data to be written either could be read sequentially or it if it was from different parts of the disk staged into a continous large file on a fast disk device, fast enough that the tape device's real time demands could be meet.
Thank you for this great journey down the memory lane.
You are welcome !
Great work, Moshix! Very informative as usual in your videos. I also think many of us we would like to know more about that recent acquisition :)
I worked with 3370 and 3380 drives in places I worked. Funny thing about the 3370, some of those would make a sound like a buzz saw once in a while, did they ever break down, uh nope. Once saw a 3380 catch fire, it was the main system volume for VM/370, did VM/370 crash, nope, it simply said the volume was offline, when we hit the big red switch. That system continued to run until 8pm, when the system was shutdown, the backup tape loaded it on another volume, performed an IPL, and the next day the techs were on-site to repair the drive.
Wow. Amazing that it continued to run!
Never had the chance to work with mainframes but when I studied to be ready if the chance arose I was told big iron systems have so much redundancy one could claim you could physically disable or destroy half of it and it would still keep running in some capacity.
Wish Windows and sometimes even Linux had that level of fault tolerance, a literal lp0 printer on fire scenario without crashing or major disruption
@Mordecrox indeed
VM sysprog/fanboy here. For reliability we had a VM system that ran continuously for almost a year without shutdown. The only reason it didn't run longer was we had to apply some maintenance that required a restart.
@stan.rarick8556 and I am a VM fanboy too !
Coming from Win and Linux, I always wanted to learn about mainframe computing. This channel together with tk4/5 make a great resource. Remember that commercial „my own IBM computer. Imagine that“?
Welcome to the channel !
The IBM 1311 and 2311 were hydraulic actuated. I remember that if they were not squarely mounted on the floor the power of the actuator moving between cylinders would walk the driver across the floor., More than once we had a hydraulic line or connection fail and the device would be flooded with hydraulic fluid. 7.25 MB of storage seems minuscule in size today but it could held our compilers and programs and meant we didn't have to load them from cards every time we wanted to run them.
Great story. Thank you
The 2314 was hydraulic as well. Third party manufacturers made voice coil versions of the 2311 and 2314.
Didn’t say it wasn’t hydraulic
when I was eleven years old, the math-science center had an IBM 1620 with an attached 2311 hard drive. It was obsolete (donated by philip morris), but made great Snoopy calendar printouts. did some fortran code for that computer. lots of blinkenlights. saw an operational IBM 1401 at the defense general supply center in chesterfield VA in 1973. was briefly an IBM/360 operator in 1979 for a financial services company.
I ran a school dàta center's DOS 26.2 system on 2 2311s. Generating a new supervisor was fun because you had to do a lot of pack swapping
I was a S/370 console operator for a large insurance company that had a 3032 mainframe and a vast computer room full of these 3350s. (I was only 19 years old)
when we would start up batch processing in the evening, it was as if a hundred washing machines hit the spin cycle all at once.
Mainframe operator was my favorite job ever. Did that from '79 through '83, then became a systems programmer, i.e., installer and sysadmin.
left the mainframe world in '87 never to return.
Yes the fun we had in those days is just not available anymore. You felt that it was an important job and to be taken seriously but it was still a lot of fun
@@moshixmainframechannel I was 19 and making $24k / year plus full benefits with no college degree (1979). most of my friends earned half that amount at their jobs.
everyone in DP operations in the town I lived in knew everyone else. job hopping usually meant a 15% raise.
I do recall that everyone working 3rd shift everywhere I ever worked was C R A Z Y - mostly in a good way.
the execs never had any idea what was going on after midnight behind locked doors in the machine room . . . .
TOTALLY ...and the tape drives right behind,, acting like banks of clothes dryers. It definitely was like working at the Bizarro Jetsons Laundromat. Tape Ape ,and proud of it...
@@bzert281 pulling scratch tapes - - - UGH, I hated that duty - but it wasn't as bad as decollating 4-part carbon fanfold paper
You left the mainframe world to which other world?
Just retired. Started work in IT in 1976 on an IBM S360/40 Have used all these DASD systems except 3340. I remember when our site reached 1.4TB of online DASD. (3380's & 3390's)
Sounds like a Canadian guy I know…. I will retire soon after 47 years. Cheers Paul
This stuff takes me down memory lane. Moshix would like the unofficial museum in POK
I would
For quite some time in the late 1980s, IBM subcontracted manufacturing their disk arrays to Hitachi Computer in Kanagara, Japan. They were able to produce them with only a one micron head to disk clearance which used less power and had faster read write speeds. IBM then licensed the tech from Hitachi and then manufactured their own.
Yes, I had that rumor. Nice to know the rumor was true.
I don't think the timeline is right. The IBM General Products Division in San Jose, CA, manufactured the disk equipment. Their first RAID disk product was called the RAMAC (reuse of the 1950s disk name), announced in June of 1994, and was manufactured by IBM competitor Storage Technology Corp for IBM after EMC Corporations RAID disk products cut deeply into IBM's sales. IBM announced the sale of its Storage Technology Division to Hitachi in 2002.
@@jpcallan97225 ok but the title of the video is about S/360 disk devices. Ramac was before S/360
@@moshixmainframechannel You are correct about the 305 Ramac predating System/360, but IBM reused the Ramac name, rushing a repackaged competitor's product to market. Their first RAID disk product was called the RAMAC (reuse of the 1950s 305 disk name), announced in June of 1994, and was manufactured for IBM by competitor Storage Technology Corp after EMC Corporations RAID disk products cut deeply into IBM's DASD sales. I don't remember the 1994 Ramac's machine number. I didn't like the product and was recommending EMC's cached RAID arrays to my consulting clients.
I used them in the middle 80's they were used in Buenos Aires University, to store Student records. We called them sandwicheras ( sandwich storage ), because they were pretty much like the ones used in bars.I still remember the VOL names "BAGUAL" and "MANIJA", and the slang "to scratch a disk"
Feldman!! Shalom !
Very nice video with good technical info and slides. This would also be a good video to use in a computer science course to educate someone on the history of this media. Nicely done.
Thank you!
History: First removal disk pack 1311(went on 1440 computers. For 360 ibm renumbered 1311 changed to 2311 disk drive.(really same drive, higher density). Winchester drives
were developed at IBM lab located in winchester england. 3330 disk packs cost $1,000
per pack(100 meg pack!!!!).
I worked with IBM, as a language instructor, at the Hursley Park Labs during the time the Winchester Disk Systems were designed and developed (in the 1960's). The programming language PL/1 was also developed at that location and I played a part in writing some compiler routines for that much denigrated development tool. Those were great, exciting, days to work at IBM when the company, headed by Tom Watson Jr, was on track to become the world's biggest company. Where did it all go wrong?
@jimhardy2034 exciting experience !! Thanks for sharing. Where it went wrong? Easy: IBM always had an attitude that doesn’t work anymore nowadays: i.e. that they know better than the customer what the customer needs. That’s what’s wrong with IBM. In other words, arrogance
I'm pretty sure that drums were used as I/O devices. I remember taking a trip to the American Airlines reservations system in the '60s and gazing at their 7330 drum attached to their 7090. It's, essentially, a head-per-track disk, no seek time. Also, all the virtual memory 360s and 370s pretty much had to have 2301 or 2303 drums for paging devices if they ran CP67 or VM. Unlike HyperV (somebody correct me if I'm wrong) CP and VM can page the virtual machines they create not keep the whole virtual machine in memory.
HyperV pages guest VM memory
On the 360 drums were channel attached and used as I/O devices. Some early drums used drums for main memory. The 7090 used core memory, so indeed the drum was an I/O device.
I believe that the System Managed Storage feature provides the badly needed abstractin needed to provde potability to keep up with technology. It provides huge improvement in JCL simplification. Also, with move towards database systems like DB2, Oracle and My SQL, separation from the nature of the physical device record layout is complete. Whenever the data type of an item is changed, ,added or remvoed or archived, the systems will deal without it without extra programming!
Thanks for writing. In theory yes, in practice it created the need for a storage admin, which was not a role that existed prior to SMS
@@moshixmainframechannel I never had the opportunity to learn what items that were the opbject or storage administration setup. When I was programming for apps on the Prime supermini systems, we were never exposed to the nature of device physical records and how many of the physical records were contained on a device track!
did you mention that an HLQ had a max length of 8 characters? I may have missed it.
8.3 filenames, meet 8.8.8.8.8 filenames, LOL
Lovely history. I was "born" into mainframes in 1980, we timeshared, and had two 3350 packs assigned for our use. When we in-housed 2 years later, we bought a 4381, four 3420 tape drives, two 3380s, a 3705 NCP, and a 2501 card reader (yes a card shop!). we also rented two 3350s for about a year, to help us migrate our jobs. First CLIST i think i wrote, was, drum roll, a blocksize calculator (hello, world!), and yeah, those 3350 numbers stick with you. When i'm old and demented, i'll probably still remember 555 x 10 x 19069, even though it's been 39 years! My first sysgen was a CBIPO of MVS/SP for that move - and a PEPgen for the RJE remotes and dial-out stuff.
Hi Moshix, love your channel, great fun playing with these vintage systems on my windows PC (even though I work on various z/OS systems professionally it's difficult to play with things outside my specialisms). I was wondering - whether you have considered doing a video on setting up applications that use started tasks. Is anything available to be downloaded? (I know there is KICKS, but I'm unsure whether that can be set up to run as an STC in MVS3.8 in the way CICS can under z/OS). Or, is it possible to create a simple user-defined app that runs as an STC under MVS3.8? I'm just thinking of the basics of how to define the JCL/procudures, stopping/starting tasks, maybe a bit of automation, APF authorisations etc. Anyway keep up the great work really appreciate the videos, always dense with history and knowledge to learn from.
Hey Moshix, a topic for future consideration - I'd really like to get into the basics of interactive mainframe apps on MVS 3.8 and the available languages and tools we could use today to develop interactive panels, including menus and data entry forms. I know you've covered the basics of installing KICKS in the past and I'm about to try that myself; does REXX for MVS 3.8 also have the capability to author panels, and what other tools could we consider, hack around with, study from, or learn about, for example how was the TK4- menu system developed?
Well with the new BREXX 2.3 it’s easy to build panels in TSO in MVS3.8. But you would have to wait for Update 9 of MVS 3.8 TK4. Hopefully out by the end of the year.
The good old DS6800 as photographed for the present day. However even though I have one I will be restoring I would never trust one in production. The main reason is the fact of the KONA cards fail at random without warning and if I properly quiesced they will sometimes spontaneously format. But anyway I would trust my data more to a DS8K line of hardware.
Yea the DS6800 is the worst mainframe product IBM made these last 20 years. No doubt
Yes which is why I’m looking into dumping it to tape and transferring it to a DS8800 that was decommissioned from the Miami university. The system was running 2 VSE LPAR in ESA/390 mode. Only one issue is I need to buy replacements disks for the DS8800 and the CECs has the disks removed but I believe the previous CE left the restore media so I’m looking into restoring it. However when I’m all said and done if anyone wants that DS6k with CKD Licence I am looking into selling it everything disks software and expansion for the price of 1 KONA.
Hi moshix, not related to this video, but I was hoping you could point me to the best IBM Assembly Language reference (in PDF form) which runs under MVS 3.8j. I found some on bitsavers but was hoping you could name "the one" you like most. Thanks!
You skipped the 2310, which is the single-platter removable disk (the platter was the 2315), developed for the 1130 and 1800 (where it was a/k/a 1810) but also used in the 360/44. DEC adopted the drive as the RK-05.
You’re right !
DASD, pronounced "daz dee" is direct access storage device... I think you might have said attached. Anyway fantastic video as always but I've particularly enjoyed your hardware related stories. You might want to investigate other CPU devices such as console devices and even the motor generator sets the accompanied processors from at least as early as the 135 through to the 4300 series, converting 50/60 Hz to 400 Hz. The were typically used as tables in machine rooms as they were about the same size as a 4381 CPU.
Nice video! I learned a lot.
Dataset Plisource Qualifier has 9 char
22:50 your photo is not a 3330. The 3330 was a pizza oven configuration like the 2314. The model 1 and 11 packs were different and not interchangeable.
Didn’t say they were interchangeable
You image of the 3330 is incorrect it is actually a 3370/3375. The 3330 is shown on the previous slide. I am a retired IBM CE
Thx
quick question: what does the H stand for in HNET? :)
Hercules
Great work
Senkiu
How much did the disk packs cost?
I don’t know. Good question
Awesome video! I love it!
I noticed your channel name changed to "moshix moshix"
Any reason behind that?
is the unit a SMD ?
What’s an smd ?
@@moshixmainframechannel System Mounted Drive not sure a system used by Quentel Paint Box to attaching storage devices.
I noticed that above the Title of this video it says Whidbey Island. Any reason why?
Yes I was there when I made the video
Sorry for my late reply but I was stationed on whidbey Island when I was in the Navy.
Oh lovely!
"virtual memory wouldn't exist without fast disk" - wellllllllllll, it probably WOULD, because it was so much of a bargain to be able to swap work in-and-out-of-memory, and run "more work than your CPU could run" - but it would have been less performance, definitely. we'd be looking at hours of backlogs that we didn't have to look at with fast drives.... but you know how "money talks," and if the company didn't have to buy more capacity, they'd put up with things being slower...within reason.
(p.s. at least MY company would make that compromise --- but they were cheapskates! ;)
Okay but where would it swap to, if no disk is available ?
@@moshixmainframechannel well, if no disks were available at all, then that would be gruesome indeed. I didn't think you meant, that, just that a shop might not allocate fastest disk. Fortunately it's an academic exercise we never had to try, but i always wondered about the possibility of having a tape or two permanently mounted, (just for swaps, not for paging of-course,) if you had to extend memory in a disk-free system, and could tolerate a region-swap taking a minute or maybe more. But, now I think of it, I don't believe ASM has the methods available, to manage a set of sequential-media "swap files" anyway, so scratch that idea, heh heh.
Swapping to tape did indeed exist. But those were not virtual memory address spaces. It was just memory segments. ICTS on the 7094 is an example of that
I worked at a shop in 1967 where we rolled a BG partition out to tape, loaded a transient program in from the same tape, rewound the tape while the transient program was running, then reloaded the original BG batch program back in to continue execution...........a form of virtual memory on tape. IBM did something similar but to disk IIRC
LINK ACC 192
There are so many errors in here that I don't even know where to start. My credentials are 46 years as an IBM mainframe administrator/programmer from 1966 on the 360/40 through the Z series and 2311 through 3390. You can see me in my icon/avatar seated at an IBM 360/44 running 44/MFT in 1967.
You can start by being nice. Jerk.
@@moshixmainframechannel Agree and I don't mind corrections in the comments, I always read the comments to see the "errata". This dude just say "Many things are wrong" but don't mention a single one, and he is harsh about nothing. Thanks for sharing the info you have, even if you have an error at least you are preserving and spreading information, for free.
@@jaimeduncan6167 thank you much !
If I'd spent 46 years of my life wrangling IBM mainframes I'd be cranky too. I spend a couple years in VM storage management at the start of my IBM career.
@@billkissick6268 hahaha true
Thank you great explanation
Welcome !
I had to stop watching because of the ambient quasi-music.
Sure. Thanks for trying