|
So I am studying the OS JCL and Utilities book (Michael Trombetta) from the mid 80's, good book ,was recommended to me for the MVS 3.8j era of JCL. Am happy with it and can also recommend it.
They discuss utilization of the MSS storage device(s) a 3850, and at the mid 80's of the book it was a current device, seems to consist of a 3330 disk (maybe more than one) and a series of tape carts, with an "invisible" autoloader system, so as to implement "near-line" storage. Only one actual dasd and a bunch of different images of it on tape, that would be recalled by the VOLSER from MVS and loaded on the live DASD where it was then a live 3330 disk. (obviously a delay in preforming this "staging operation" as I take it was called).
Probably more complicated in reality but that is my gist of the functionality. Different logically but easier than mounting a tape, doing a restore, then archiving, not as fast as having dedicated DASD.
I assume since I can't find it in Hercules, that this device and not sure to call it tape or disk - is not emulated by Hercules, probably would not be useful and no one uses any such scheme since the mid 80 As the need probably got replaced quickly by vastly larger amounts of cheap DASD being available? Anyone know? Or remember these things and why they are obscure today?
I do not see this as being the same functionality as a backup tape library (automounter) of true tapes treated as tapes which is for sure still being utilized.
Is my assessment of what this 3850 MSS device was correct? And why we don't have an emulation for such a thing anymore?
Thanks,
Dave N8ZFM
|
The 3850 (and the 2821 before it, which performed a similar function in a much different manner) were notoriously unreliable, with serious issues surrounding the function of retrieving and storing of cartridges. When they worked, they worked fairly well, but the problems were such that people tended to design their systems not to need that functionality. Only a few of the largest customers had them, and only for uses where they needed the performance. Eventually, software such as DFHSM and equivalents from Sterling Software and FDR provided the same function with standard tapes.
On Fri, Feb 7, 2025 at 12:21?AM Dave Trainor - N8ZFM via <dave= [email protected]> wrote: So I am studying the OS JCL and Utilities book (Michael Trombetta) from the mid 80's, good book ,was recommended to me for the MVS 3.8j era of JCL.? Am happy with it and can also recommend it.
They discuss utilization of the MSS storage device(s) a 3850, and at the mid 80's of the book it was a current device, seems to consist of a 3330 disk (maybe more than one) and a series of tape carts, with an "invisible" autoloader system, so as to implement "near-line" storage.? ?Only one actual dasd and a bunch of different images of it on tape, that would be recalled by the VOLSER from MVS and loaded on the live DASD where it was then a live 3330 disk. (obviously a delay in preforming this "staging operation" as I take it was called).
Probably more complicated in reality but that is my gist of the functionality.? ?Different logically but easier than mounting a tape, doing a restore, then archiving, not as fast as having dedicated DASD.
I assume since I can't find it in Hercules, that this device and not sure to call it tape or disk - is not emulated by Hercules, probably would not be useful and no one uses any such scheme since the mid 80 As the need probably got replaced quickly by vastly larger amounts of cheap DASD being available?? ? Anyone know?? ?Or remember these things and why they are obscure today?
I do not see this as being the same functionality as a backup tape library (automounter) of true tapes treated as tapes which is for sure still being utilized.
Is my assessment of what this 3850 MSS device was correct? And why we don't have an emulation for such a thing anymore?
Thanks,
Dave N8ZFM
|
On 07/02/2025 06:37, Jay Maynard via
groups.io wrote:
The 3850 (and the 2821 before it, which performed a
similar function in a much different manner) were notoriously
unreliable, with serious issues surrounding the function of
retrieving and storing of cartridges. When they worked, they
worked fairly well, but the problems were such that people
tended to design their systems not to need that functionality.
Only a few of the largest customers had them, and only for uses
where they needed the performance. Eventually, software such as
DFHSM and equivalents from Sterling Software and FDR provided
the same function with standard tapes.
I would say it was hard to design systems that made efficient use of
them. I can see that some Universities and research organizations in
the UK had them. I think at Manchester University they were used for
storing data for an individual user. I think in a commercial setting
it might be hard to design a system that made efficient use of them.
As they were unreliable I imagine you would want them backed up and
I can't see how you would achieve that..
On Fri, Feb 7, 2025 at
12:21?AM Dave Trainor - N8ZFM via <dave= [email protected]>
wrote:
So
I am studying the OS JCL and Utilities book (Michael
Trombetta) from the mid 80's, good book ,was recommended to me
for the MVS 3.8j era of JCL.? Am happy with it and can also
recommend it.
They discuss utilization of the MSS storage device(s) a 3850,
and at the mid 80's of the book it was a current device, seems
to consist of a 3330 disk (maybe more than one) and a series
of tape carts, with an "invisible" autoloader system, so as to
implement "near-line" storage.? ?Only one actual dasd and a
bunch of different images of it on tape, that would be
recalled by the VOLSER from MVS and loaded on the live DASD
where it was then a live 3330 disk. (obviously a delay in
preforming this "staging operation" as I take it was called).
Probably more complicated in reality but that is my gist of
the functionality.? ?Different logically but easier than
mounting a tape, doing a restore, then archiving, not as fast
as having dedicated DASD.
Not really. Thats pretty much what it did. The complex bit probably
writing the updated disk track back to the 3330.
I assume since I can't find it in Hercules, that this device
and not sure to call it tape or disk - is not emulated by
Hercules, probably would not be useful and no one uses any
such scheme since the mid 80 As the need probably got replaced
quickly by vastly larger amounts of cheap DASD being
available?? ? Anyone know?? ?Or remember these things and why
they are obscure today?
Not many made. Really only useful in special, e.g. R&D settings.
Quickly eclipsed by cartridge loaders and larger disks. Some
recollections by VMers are here
I do not see this as being the same functionality as a backup
tape library (automounter) of true tapes treated as tapes
which is for sure still being utilized.
It wasn't. A common use was in Universities and research where each
user had a cartridge.
Is my assessment of what this 3850 MSS device was correct?
Pretty much
And
why we don't have an emulation for such a thing anymore?
I think no love for it. People write emulations because they have a
love for something. No one loved the MSS.
Also getting to right amount of unreliability to make a realistic
emulation might be tricky, even on windows.
--
Dave
G4UGM
|
toggle quoted message
Show quoted text
On Fri, Feb 7, 2025 at 12:21?AM Dave Trainor - N8ZFM via <dave= [email protected]> wrote: So I am studying the OS JCL and Utilities book (Michael Trombetta) from the mid 80's, good book ,was recommended to me for the MVS 3.8j era of JCL.? Am happy with it and can also recommend it.
They discuss utilization of the MSS storage device(s) a 3850, and at the mid 80's of the book it was a current device, seems to consist of a 3330 disk (maybe more than one) and a series of tape carts, with an "invisible" autoloader system, so as to implement "near-line" storage.? ?Only one actual dasd and a bunch of different images of it on tape, that would be recalled by the VOLSER from MVS and loaded on the live DASD where it was then a live 3330 disk. (obviously a delay in preforming this "staging operation" as I take it was called).
Probably more complicated in reality but that is my gist of the functionality.? ?Different logically but easier than mounting a tape, doing a restore, then archiving, not as fast as having dedicated DASD.
I assume since I can't find it in Hercules, that this device and not sure to call it tape or disk - is not emulated by Hercules, probably would not be useful and no one uses any such scheme since the mid 80 As the need probably got replaced quickly by vastly larger amounts of cheap DASD being available?? ? Anyone know?? ?Or remember these things and why they are obscure today?
I do not see this as being the same functionality as a backup tape library (automounter) of true tapes treated as tapes which is for sure still being utilized.
Is my assessment of what this 3850 MSS device was correct? And why we don't have an emulation for such a thing anymore?
Thanks,
Dave N8ZFM
-- Mike A Schwab, Springfield IL USA
Where do Forest Rangers go to get away from it all?
|
Dave,
The 3850 mass store was at the time quite a device -
And due to the high cost of DASD at the time - it really did serve a purpose -
At the larger shops that had these - you will hear all manner of horror stories - there were some doozies - believe me -
The device was of course very mechanical - and - as such - had many places and opportunities to have failures -
This was of early introduction of robotics to do the mounts -
THANKFULLY they are gone - relegated to these old stories -
Today - we use similar techniques - with HSM (Hierarchical Storage Manager) to back up and recall when required data sets -
And of course, HSM has evolved - from using special staging disks to migrating to tape - with auto mounts - and now in today's world VTS which are now entire virtual tape systems - that are entire units that are full of staging DASD - and special units that EMULATE tape -
Remember - IBM no longer even sells spinning rust - everything is now solid state - shows how much the technology has evolved.
Regards
-J-
Jeff Bassett
Bassettj@...
(301) 424-3362 (office)
(240) 388-7148 Cell
Time spent flying - is NOT deducted from one’s lifespan
toggle quoted message
Show quoted text
-----Original Message----- From: [email protected] < [email protected]> On Behalf Of Dave Trainor - N8ZFM via groups.io Sent: Friday, February 7, 2025 1:22 AM To: [email protected]Subject: [H390-MVS] Historical Question So I am studying the OS JCL and Utilities book (Michael Trombetta) from the mid 80's, good book ,was recommended to me for the MVS 3.8j era of JCL. Am happy with it and can also recommend it. They discuss utilization of the MSS storage device(s) a 3850, and at the mid 80's of the book it was a current device, seems to consist of a 3330 disk (maybe more than one) and a series of tape carts, with an "invisible" autoloader system, so as to implement "near-line" storage. Only one actual dasd and a bunch of different images of it on tape, that would be recalled by the VOLSER from MVS and loaded on the live DASD where it was then a live 3330 disk. (obviously a delay in preforming this "staging operation" as I take it was called). Probably more complicated in reality but that is my gist of the functionality. Different logically but easier than mounting a tape, doing a restore, then archiving, not as fast as having dedicated DASD. I assume since I can't find it in Hercules, that this device and not sure to call it tape or disk - is not emulated by Hercules, probably would not be useful and no one uses any such scheme since the mid 80 As the need probably got replaced quickly by vastly larger amounts of cheap DASD being available? Anyone know? Or remember these things and why they are obscure today? I do not see this as being the same functionality as a backup tape library (automounter) of true tapes treated as tapes which is for sure still being utilized. Is my assessment of what this 3850 MSS device was correct? And why we don't have an emulation for such a thing anymore? Thanks, Dave N8ZFM
|
When I worked for IBM in Tucson, AZ, in the early 1980's, much of our data was stored on MSS. When we would run a foreground job from TSO, the application would typically output a message like: "Allocating datasets, please be patient". That is because executing a statement like "ALLOC DD(xxxx) DSN(xxx.yyy.zzz)" took a half minute to a full minute to execute. The MSS monstrosity was amazing to watch in operation. It consisted of a wall of honeycomb-like compartments where the tapes were stored. The operation of the whole contraption was mostly transparent to the user. The command to allocate a dataset would cause the picker to locate the honeycomb location where the tape was stored, retrieve it, and mount it on a device that emulated a DASD. I don't think the tape was copied to a real DASD. Data was read from and written directly to the tape, which was quite wide. I think each vertical (crosswise) stripe on the tape emulated a DASD track. When a FREE command was executed, the tape would be rewound back into the cartridge and the picker would put it back into the honeycomb. I'm not sure if any special support would be required from Hercules. I think the VOLSER of the tape appeared to the OS as just another 3330 DASD. (This is all from my distant memory of conversations with older engineers would used to work on the beast. I'm sure an internet search would yield a much better description.)
Charles Bailey
toggle quoted message
Show quoted text
On 2025-02-07 00:21, Dave Trainor - N8ZFM wrote: So I am studying the OS JCL and Utilities book (Michael Trombetta) from the mid 80's, good book ,was recommended to me for the MVS 3.8j era of JCL. Am happy with it and can also recommend it.
They discuss utilization of the MSS storage device(s) a 3850, and at the mid 80's of the book it was a current device, seems to consist of a 3330 disk (maybe more than one) and a series of tape carts, with an "invisible" autoloader system, so as to implement "near-line" storage. Only one actual dasd and a bunch of different images of it on tape, that would be recalled by the VOLSER from MVS and loaded on the live DASD where it was then a live 3330 disk. (obviously a delay in preforming this "staging operation" as I take it was called).
Probably more complicated in reality but that is my gist of the functionality. Different logically but easier than mounting a tape, doing a restore, then archiving, not as fast as having dedicated DASD.
I assume since I can't find it in Hercules, that this device and not sure to call it tape or disk - is not emulated by Hercules, probably would not be useful and no one uses any such scheme since the mid 80 As the need probably got replaced quickly by vastly larger amounts of cheap DASD being available? Anyone know? Or remember these things and why they are obscure today?
I do not see this as being the same functionality as a backup tape library (automounter) of true tapes treated as tapes which is for sure still being utilized.
Is my assessment of what this 3850 MSS device was correct? And why we don't have an emulation for such a thing anymore?
Thanks,
Dave N8ZFM
|
Charles,
All true - with the caveat - when it worked - it was fun and amazing - HOWEVER, when it failed - there were some horrific messes -
We used to laugh that this was a VERY INTELLIGENT MACHINE - such, that when it would discover that there was some error condition on a cartridge - it would IMMEDIATELY grab it and DESTROY IT - so you would not have to deal with bad data - HA
-J-
Jeff Bassett
Bassettj@...
(301) 424-3362 (office)
(240) 388-7148 Cell
Time spent flying - is NOT deducted from one’s lifespan
toggle quoted message
Show quoted text
-----Original Message----- From: [email protected] < [email protected]> On Behalf Of Charles Bailey via groups.io Sent: Friday, February 7, 2025 11:23 AM To: [email protected]Subject: Re: [H390-MVS] Historical Question When I worked for IBM in Tucson, AZ, in the early 1980's, much of our data was stored on MSS. When we would run a foreground job from TSO, the application would typically output a message like: "Allocating datasets, please be patient". That is because executing a statement like "ALLOC DD(xxxx) DSN(xxx.yyy.zzz)" took a half minute to a full minute to execute. The MSS monstrosity was amazing to watch in operation. It consisted of a wall of honeycomb-like compartments where the tapes were stored. The operation of the whole contraption was mostly transparent to the user. The command to allocate a dataset would cause the picker to locate the honeycomb location where the tape was stored, retrieve it, and mount it on a device that emulated a DASD. I don't think the tape was copied to a real DASD. Data was read from and written directly to the tape, which was quite wide. I think each vertical (crosswise) stripe on the tape emulated a DASD track. When a FREE command was executed, the tape would be rewound back into the cartridge and the picker would put it back into the honeycomb. I'm not sure if any special support would be required from Hercules. I think the VOLSER of the tape appeared to the OS as just another 3330 DASD. (This is all from my distant memory of conversations with older engineers would used to work on the beast. I'm sure an internet search would yield a much better description.) Charles Bailey On 2025-02-07 00:21, Dave Trainor - N8ZFM wrote: So I am studying the OS JCL and Utilities book (Michael Trombetta) from the mid 80's, good book ,was recommended to me for the MVS 3.8j era of JCL. Am happy with it and can also recommend it.
They discuss utilization of the MSS storage device(s) a 3850, and at the mid 80's of the book it was a current device, seems to consist of a 3330 disk (maybe more than one) and a series of tape carts, with an "invisible" autoloader system, so as to implement "near-line" storage. Only one actual dasd and a bunch of different images of it on tape, that would be recalled by the VOLSER from MVS and loaded on the live DASD where it was then a live 3330 disk. (obviously a delay in preforming this "staging operation" as I take it was called).
Probably more complicated in reality but that is my gist of the functionality. Different logically but easier than mounting a tape, doing a restore, then archiving, not as fast as having dedicated DASD.
I assume since I can't find it in Hercules, that this device and not sure to call it tape or disk - is not emulated by Hercules, probably would not be useful and no one uses any such scheme since the mid 80 As the need probably got replaced quickly by vastly larger amounts of cheap DASD being available? Anyone know? Or remember these things and why they are obscure today?
I do not see this as being the same functionality as a backup tape library (automounter) of true tapes treated as tapes which is for sure still being utilized.
Is my assessment of what this 3850 MSS device was correct? And why we don't have an emulation for such a thing anymore?
Thanks,
Dave N8ZFM
|
Even modern machines can be interesting. I was on-site at an implementation of a new IBM tape robot box a few years back. The box tested out fine, but intermittently shut down when we were loading tapes. Eventually during one shutdown one robot broke its plastic fingers off.
It turned out one of the doors had a problem with its sensor switch. The movement of loading the tapes caused the machine to think the door had opened, and to protect puny humans the arm was programmed to immediately drop to the floor. By chance, that occurred while the robot had its fingers in a tape slot. It was willing to harm itself instead of possibly harm a human.
toggle quoted message
Show quoted text
On 2/7/2025 8:27 AM, Jeff Bassett wrote: Charles,
All true - with the caveat - when it worked - it was fun and amazing - HOWEVER, when it failed - there were some horrific messes -
We used to laugh that this was a VERY INTELLIGENT MACHINE - such, that when it would discover that there was some error condition on a cartridge - it would IMMEDIATELY grab it and DESTROY IT - so you would not have to deal with bad data - HA
-J-
Jeff Bassett
Bassettj@...
(301) 424-3362 (office)
(240) 388-7148 Cell
Time spent flying - is NOT deducted from one’s lifespan
-----Original Message-----
From: [email protected] <[email protected]> On Behalf Of Charles Bailey via groups.io
Sent: Friday, February 7, 2025 11:23 AM
To: [email protected]
Subject: Re: [H390-MVS] Historical Question
When I worked for IBM in Tucson, AZ, in the early 1980's, much of our data was stored on MSS. When we would run a foreground job from TSO, the application would typically output a message like: "Allocating datasets, please be patient". That is because executing a statement like "ALLOC DD(xxxx) DSN(xxx.yyy.zzz)" took a half minute to a full minute to execute. The MSS monstrosity was amazing to watch in operation. It consisted of a wall of honeycomb-like compartments where the tapes were stored. The operation of the whole contraption was mostly transparent to the user. The command to allocate a dataset would cause the picker to locate the honeycomb location where the tape was stored, retrieve it, and mount it on a device that emulated a DASD. I don't think the tape was copied to a real DASD. Data was read from and written directly to the tape, which was quite wide. I think each vertical (crosswise) stripe on the tape emulated a DASD track. When a FREE command was executed, the tape would be rewound back into the cartridge and the picker would put it back into the honeycomb. I'm not sure if any special support would be required from Hercules. I think the VOLSER of the tape appeared to the OS as just another 3330 DASD.
(This is all from my distant memory of conversations with older engineers would used to work on the beast. I'm sure an internet search would yield a much better description.)
Charles Bailey
On 2025-02-07 00:21, Dave Trainor - N8ZFM wrote:
So I am studying the OS JCL and Utilities book (Michael Trombetta) from the mid 80's, good book ,was recommended to me for the MVS 3.8j era of JCL. Am happy with it and can also recommend it.
They discuss utilization of the MSS storage device(s) a 3850, and at the mid 80's of the book it was a current device, seems to consist of a 3330 disk (maybe more than one) and a series of tape carts, with an "invisible" autoloader system, so as to implement "near-line" storage. Only one actual dasd and a bunch of different images of it on tape, that would be recalled by the VOLSER from MVS and loaded on the live DASD where it was then a live 3330 disk. (obviously a delay in preforming this "staging operation" as I take it was called).
Probably more complicated in reality but that is my gist of the functionality. Different logically but easier than mounting a tape, doing a restore, then archiving, not as fast as having dedicated DASD.
I assume since I can't find it in Hercules, that this device and not sure to call it tape or disk - is not emulated by Hercules, probably would not be useful and no one uses any such scheme since the mid 80 As the need probably got replaced quickly by vastly larger amounts of cheap DASD being available? Anyone know? Or remember these things and why they are obscure today?
I do not see this as being the same functionality as a backup tape library (automounter) of true tapes treated as tapes which is for sure still being utilized.
Is my assessment of what this 3850 MSS device was correct? And why we don't have an emulation for such a thing anymore?
Thanks,
Dave N8ZFM
|
Oh yes .. the MSS (or Mess as we used to call it). The JES3 interfaces were horrendous - JES3 tried to own all tapes and disks. The cobbled together MSS support was a support nightmare. At the time I was supporting one of our JES3 complexes. ISTR we had 2 of the beasts and they were heavy - the floor had to be checked for weight limits. I used to have a cartridge somewhere but it got lost in one of my moves (or my wife threw it out, along with a complete set of the PL/1 Optimizing and Checkout compiler manuals). That and the IBM tape library were 2 of the not so great ideas from big blue. The STC tape library (the round one) was faster, had better support.
|
Back in, probably the mid 1990's, the competitive analysis team of
the tape development mission in Tucson, bought one of these STK tape
library silos and played around with it.? They wanted to find a way
to increase market share for our IBM tape drives.? They developed a
modified version of our tape drive's microcode so that it would
identify itself as an STK tape drive and accept commands from the
STK silo.? This would allow it to be mounted in an STK silo in place
of an STK drive.
IBM's first offering of a tape library was a fast-track effort to
throw together something that could compete with STK's tape
library.? They re-purposed some giant robot mechanism that had been
developed for something else to make it do the job of storing and
retrieving tapes from a long linear library.? They nicknamed it
"Conan the Librarian".? It was a huge overkill but it did the job.
IBM's later tape libraries were much better.? They had several of
these in some of our labs running 24/7 with some software that would
issue continuous random requests for mounts and unmounts of tapes,
looking for failures so that the cause could be investigated and
corrected.? I think some of the largest libraries could hold 50K
tape cartridges.
Charles Bailey
On 2025-02-07 10:44, S. L. Garwood via
groups.io wrote:
toggle quoted message
Show quoted text
Oh yes .. the MSS (or Mess as we used to call it). The JES3
interfaces were horrendous - JES3 tried to own all tapes and
disks. The cobbled together MSS support was a support nightmare.
At the time I was supporting one of our JES3 complexes. ISTR we
had 2 of the beasts and they were heavy - the floor had to be
checked for weight limits. I used to have a cartridge somewhere
but it got lost in one of my moves (or my wife threw it out,
along with a complete set of the PL/1 Optimizing and Checkout
compiler manuals). That and the IBM tape library were 2 of the
not so great ideas from big blue. The STC tape library (the
round one) was faster, had better support.
|
This discussion has been interesting, a good idea that never really worked out due to implementation. --- I was unaware that IBM had discontinued all spinning platters, while I certainly myself would not think of using anything other than a SSD as my boot device on any machine. It seems to me there is still a need for the larger spinning disks in array's for massive amounts of data, would be dog gone expensive to put all that on solid state and the speed is not required in that application, like for my video storage of movies and tv shows. But I guess as time goes on it for sure will be the future. Thanks for the interesting stories of this device I had not heard of before, seems like I was lucky I missed out on the 3850. Thanks, Dave ?On 2/7/25, 11:03 AM, " [email protected] <mailto: [email protected]> on behalf of Jeff Bassett" < [email protected] <mailto: [email protected]> on behalf of bassettj@... <mailto:bassettj@...>> wrote: Dave, The 3850 mass store was at the time quite a device - And due to the high cost of DASD at the time - it really did serve a purpose - At the larger shops that had these - you will hear all manner of horror stories - there were some doozies - believe me - The device was of course very mechanical - and - as such - had many places and opportunities to have failures - This was of early introduction of robotics to do the mounts - THANKFULLY they are gone - relegated to these old stories - Today - we use similar techniques - with HSM (Hierarchical Storage Manager) to back up and recall when required data sets - And of course, HSM has evolved - from using special staging disks to migrating to tape - with auto mounts - and now in today's world VTS which are now entire virtual tape systems - that are entire units that are full of staging DASD - and special units that EMULATE tape - Remember - IBM no longer even sells spinning rust - everything is now solid state - shows how much the technology has evolved. Regards -J- Jeff Bassett Bassettj@... <mailto:Bassettj@...> (301) 424-3362 (office) (240) 388-7148 Cell Time spent flying - is NOT deducted from one’s lifespan
-----Original Message----- From: [email protected] <mailto: [email protected]> < [email protected] <mailto: [email protected]>> On Behalf Of Dave Trainor - N8ZFM via groups.io Sent: Friday, February 7, 2025 1:22 AM To: [email protected] <mailto: [email protected]> Subject: [H390-MVS] Historical Question So I am studying the OS JCL and Utilities book (Michael Trombetta) from the mid 80's, good book ,was recommended to me for the MVS 3.8j era of JCL. Am happy with it and can also recommend it. They discuss utilization of the MSS storage device(s) a 3850, and at the mid 80's of the book it was a current device, seems to consist of a 3330 disk (maybe more than one) and a series of tape carts, with an "invisible" autoloader system, so as to implement "near-line" storage. Only one actual dasd and a bunch of different images of it on tape, that would be recalled by the VOLSER from MVS and loaded on the live DASD where it was then a live 3330 disk. (obviously a delay in preforming this "staging operation" as I take it was called). Probably more complicated in reality but that is my gist of the functionality. Different logically but easier than mounting a tape, doing a restore, then archiving, not as fast as having dedicated DASD. I assume since I can't find it in Hercules, that this device and not sure to call it tape or disk - is not emulated by Hercules, probably would not be useful and no one uses any such scheme since the mid 80 As the need probably got replaced quickly by vastly larger amounts of cheap DASD being available? Anyone know? Or remember these things and why they are obscure today? I do not see this as being the same functionality as a backup tape library (automounter) of true tapes treated as tapes which is for sure still being utilized. Is my assessment of what this 3850 MSS device was correct? And why we don't have an emulation for such a thing anymore? Thanks, Dave N8ZFM
|
On Fri, Feb 07, 2025 at 11:00:41PM +0000, Dave Trainor - N8ZFM wrote: It seems to me there is still a need for the larger spinning disks in
array's for massive amounts of data, would be dog gone expensive to put
all that on solid state and the speed is not required in that application,
like for my video storage of movies and tv shows. My current employer recently received three new DS8A00 storage arrays, the latest, 10th generation, in IBM's DS8000 SAN storage array line. All drives in the arrays are flash core modules. -- Kevin Bruceville, TX What's the definition of a legacy system? One that works! Errare humanum est, ignoscere caninum.
|
That is certainly better than the reputation the MSS had; when the Central Bank switched to IBM, we had NCR CRAM units ("double drop!") and at some time the MSS was discussed briefly. But rumour had it that they were quite dangerous and people had died through neglecting the safeguards. Never knew whether that was an urban legend or not.
best regards,
toggle quoted message
Show quoted text
On 7 Feb 2025, at 17:51, Tom Brennan <tom@...> wrote:
Even modern machines can be interesting. ?I was on-site at an implementation of a new IBM tape robot box a few years back. ?The box tested out fine, but intermittently shut down when we were loading tapes. ?Eventually during one shutdown one robot broke its plastic fingers off.It turned out one of the doors had a problem with its sensor switch. The movement of loading the tapes caused the machine to think the door had opened, and to protect puny humans the arm was programmed to immediately drop to the floor. ?By chance, that occurred while the robot had its fingers in a tape slot. ?It was willing to harm itself instead of possibly harm a human.On 2/7/2025 8:27 AM, Jeff Bassett wrote:Charles,
All true - with the caveat - when it worked - it was fun and amazing - HOWEVER, when it failed - there were some horrific messes -
We used to laugh that this was a VERY INTELLIGENT MACHINE - such, that when it would discover that there was some error condition on a cartridge - it would IMMEDIATELY grab it and DESTROY IT - so you would not have to deal with bad data - HA
?-J-
|
Let me clarify a few things here:
User data was never read directly from or written directly to the magnetic tapes in the MSS.
There was a viertualization layer where physical disk space was allocated in 8 cylinder chunks
on the 3330(or3350) staging direves. Do not remember if you could have more than 32 (that means 64)?
virtual disks mounted on a control unit (pair), which was christened Staging Adapter in a MSS context.
So a seek to a new cylinder which is not already on any physical dasd would require mounting
a tape, requesting dasd space, reading data to dasd, repeat the channel program.?
Any data written would eventually have to be copied back to tape of course. Least recently used style.
A worst case could be that a virtual volume was demounted (KEEP'ed) and then remounted on a
different staging adapter before data was destaged to tape from the previous adapter.?
The JES3 support intended to prevent this situation but did not alwasys manage.?
The Mass Storage Controller was an addressable device of it's own. MVS (and some MSS untilities) was
sending orders (like Mount or Create a viertual volume) on this device. As far as I know it is not
emulated by Hercules and I would not argue to introduce such support either!?
I would guess that the MVS support for MSS still is there in the TK4/5 distributions of 1980-like MVS. ??
|
ISTR that JES3 had some issues if the data to be transmitted via NJE/RJE was on a MSS volume that needed to be mounted. The line would hang until everything was ready causing timeouts and line drops.?
I also clearly remember seeing a cartridge ejected (how you got one out) when the mechanism was broken - cartridge skidded 25 or 30 feet until it hit something.?
The cartridges were black, a somewhat bigger than a toilet paper tube.
Now days its all solid state stuff - no more IBM maintenance engineer with a wrench, hammer, and a scope tuning up things.?
Where is the fun in that :-)
?
|
This brings back cringing memories! We were part of the bleeding edge community and has 2 3850 MSS devices. In the pre HSM days, we actually were quite successful in implementing a migration/retrieval system for our pretty massive DASD farm. This was in the HASP/MVT days when we created the first SHARED spool capability. Here are a couple of pictures of the last piece of the honeycomb and the CE cartridge. I keep them in our den/office .
|
In case you are interested in seeing what the 3850 MSS looked like when it was operating, fetching the tapes from the honeycomb, I found these videos at this site:
?
? ?? ?
?
Note: the text is in German but you can still enjoy the videos.? :-)
?
And this page (of the same site) has a bunch of still pictures of the 3850:
?
? ? ?
?
Enjoy,
?
Mark S. Waterbury
|
In the interest of the historical pats of this hobby, having such a device available to use in Hercules might not be a bad idea, (...or maybe it is as I am speculating). It’s not an autotape library, but it's an auto dasd library of 3330's in appearance to MVS, the support for the device is apparently in the OS already, so it’s a single address and provides a selection of 3330 volumes. - Admittedly small in size by today's standards, but the allocation in the underlying Linux disk system is not overly large, the mechanical aspects would not exist as it's all software as a Hercules device. So the issues relating to this at the time being cumbersome and prone to breakage were all in the mechanical aspect, and it would no doubt be faster in operation. Now I have no idea the effort required to create such a device, and assume one would need to have allocated the entire disk space at the time of creation of this MSS device when adding it to a config file. But it might be interesting to have. The whole allocation is not that excessively large given our host system disk space available. I mean seriously, a Tb is trivial anymore. What are the thoughts of others? I would not be capable at the moment of creating this addition to Hercules itself to allow it to be allocated, and that may be a huge task for all I know it is. This is a question to the developers of Hercules and they no doubt have many competing priorities, this would be low priority. But it might be nice to have it none the less, as a historical artifact, still useful, ecpically since the mechanical issues would not exist. For those using MVS 3.8. It truly was a good idea at the time, and of course disk sizes and technology has passed it by, but so has technology passed by a lot of this and we still play with it. What's anyone think? Dave, N8ZFM, Louisville, KY dave@... dave40272@... ?On 2/7/25, 11:23 AM, " [email protected] <mailto: [email protected]> on behalf of Charles Bailey" < [email protected] <mailto: [email protected]> on behalf of txlogicguy@... <mailto:txlogicguy@...>> wrote: When I worked for IBM in Tucson, AZ, in the early 1980's, much of our data was stored on MSS. When we would run a foreground job from TSO, the application would typically output a message like: "Allocating datasets, please be patient". That is because executing a statement like "ALLOC DD(xxxx) DSN(xxx.yyy.zzz)" took a half minute to a full minute to execute. The MSS monstrosity was amazing to watch in operation. It consisted of a wall of honeycomb-like compartments where the tapes were stored. The operation of the whole contraption was mostly transparent to the user. The command to allocate a dataset would cause the picker to locate the honeycomb location where the tape was stored, retrieve it, and mount it on a device that emulated a DASD. I don't think the tape was copied to a real DASD. Data was read from and written directly to the tape, which was quite wide. I think each vertical (crosswise) stripe on the tape emulated a DASD track. When a FREE command was executed, the tape would be rewound back into the cartridge and the picker would put it back into the honeycomb. I'm not sure if any special support would be required from Hercules. I think the VOLSER of the tape appeared to the OS as just another 3330 DASD. (This is all from my distant memory of conversations with older engineers would used to work on the beast. I'm sure an internet search would yield a much better description.) Charles Bailey
On 2025-02-07 00:21, Dave Trainor - N8ZFM wrote: So I am studying the OS JCL and Utilities book (Michael Trombetta) from the mid 80's, good book ,was recommended to me for the MVS 3.8j era of JCL. Am happy with it and can also recommend it.
They discuss utilization of the MSS storage device(s) a 3850, and at the mid 80's of the book it was a current device, seems to consist of a 3330 disk (maybe more than one) and a series of tape carts, with an "invisible" autoloader system, so as to implement "near-line" storage. Only one actual dasd and a bunch of different images of it on tape, that would be recalled by the VOLSER from MVS and loaded on the live DASD where it was then a live 3330 disk. (obviously a delay in preforming this "staging operation" as I take it was called).
Probably more complicated in reality but that is my gist of the functionality. Different logically but easier than mounting a tape, doing a restore, then archiving, not as fast as having dedicated DASD.
I assume since I can't find it in Hercules, that this device and not sure to call it tape or disk - is not emulated by Hercules, probably would not be useful and no one uses any such scheme since the mid 80 As the need probably got replaced quickly by vastly larger amounts of cheap DASD being available? Anyone know? Or remember these things and why they are obscure today?
I do not see this as being the same functionality as a backup tape library (automounter) of true tapes treated as tapes which is for sure still being utilized.
Is my assessment of what this 3850 MSS device was correct? And why we don't have an emulation for such a thing anymore?
Thanks,
Dave N8ZFM
|
On 20/02/2025 05:06, Dave Trainor -
N8ZFM via groups.io wrote:
In the interest of the historical pats of this hobby, having such a device available to use in Hercules might not be a bad idea, (...or maybe it is as I am speculating).
I think its an interesting idea but ...
It’s not an autotape library, but it's an auto dasd library of 3330's in appearance to MVS, the support for the device is apparently in the OS already, so it’s a single address and provides a selection of 3330 volumes. - Admittedly small in size by today's standards, but the allocation in the underlying Linux disk system is not overly large, the mechanical aspects would not exist as it's all software as a Hercules device.
Not everyone is a linux user!
So the issues relating to this at the time being cumbersome and prone to breakage were all in the mechanical aspect, and it would no doubt be faster in operation.
You might need to slow it down. We already have issues with emulated
tape because rewind and unload is instant.
Now I have no idea the effort required to create such a device, and assume one would need to have allocated the entire disk space at the time of creation of this MSS device when adding it to a config file. But it might be interesting to have. The whole allocation is not that excessively large given our host system disk space available. I mean seriously, a Tb is trivial anymore.
It depends how you implement it. You could just use one "huge" file
and have some king of mapping logic, or one file for each cartridge
and actually copy them to tracks on an emulated 3330.
If you have one file for each cartridge you don't need to create
them all at boot up.? You can just have a cartridge number in the
name and create as needed...
?
What are the thoughts of others? I would not be capable at the moment of creating this addition to Hercules itself to allow it to be allocated, and that may be a huge task for all I know it is. This is a question to the developers of Hercules and they no doubt have many competing priorities, this would be low priority.
I think the Hercules work isn't terrible, but like you I can't do
it. There are other things I think that are more interesting....
... real TCPIP for example...
But it might be nice to have it none the less, as a historical artifact, still useful, ecpically since the mechanical issues would not exist. For those using MVS 3.8. It truly was a good idea at the time, and of course disk sizes and technology has passed it by, but so has technology passed by a lot of this and we still play with it.
What's anyone think?
I think that what was interesting about this was the mechanical
aspects. Remove those and things get a little boring. Does any one
run tape sorts...
Dave,
N8ZFM, Louisville, KY
dave@... dave40272@...
Dave
G4UGM
|
Dave,
You asked - WASTE OF TIME AND EFFORT (IMHO)
Sorry to sound negative.
-J-
Jeff Bassett
Bassettj@...
(301) 424-3362 (office)
(240) 388-7148 Cell
Time spent flying - is NOT deducted from one’s lifespan
toggle quoted message
Show quoted text
-----Original Message----- From: [email protected] < [email protected]> On Behalf Of Dave Trainor - N8ZFM via groups.io Sent: Thursday, February 20, 2025 12:07 AM To: [email protected]Subject: Re: [H390-MVS] Historical Question In the interest of the historical pats of this hobby, having such a device available to use in Hercules might not be a bad idea, (...or maybe it is as I am speculating). It’s not an autotape library, but it's an auto dasd library of 3330's in appearance to MVS, the support for the device is apparently in the OS already, so it’s a single address and provides a selection of 3330 volumes. - Admittedly small in size by today's standards, but the allocation in the underlying Linux disk system is not overly large, the mechanical aspects would not exist as it's all software as a Hercules device. So the issues relating to this at the time being cumbersome and prone to breakage were all in the mechanical aspect, and it would no doubt be faster in operation. Now I have no idea the effort required to create such a device, and assume one would need to have allocated the entire disk space at the time of creation of this MSS device when adding it to a config file. But it might be interesting to have. The whole allocation is not that excessively large given our host system disk space available. I mean seriously, a Tb is trivial anymore. What are the thoughts of others? I would not be capable at the moment of creating this addition to Hercules itself to allow it to be allocated, and that may be a huge task for all I know it is. This is a question to the developers of Hercules and they no doubt have many competing priorities, this would be low priority. But it might be nice to have it none the less, as a historical artifact, still useful, ecpically since the mechanical issues would not exist. For those using MVS 3.8. It truly was a good idea at the time, and of course disk sizes and technology has passed it by, but so has technology passed by a lot of this and we still play with it. What's anyone think? Dave, N8ZFM, Louisville, KY dave@... dave40272@... ?On 2/7/25, 11:23 AM, " [email protected] <mailto: [email protected]> on behalf of Charles Bailey" < [email protected] <mailto: [email protected]> on behalf of txlogicguy@... <mailto:txlogicguy@...>> wrote: When I worked for IBM in Tucson, AZ, in the early 1980's, much of our data was stored on MSS. When we would run a foreground job from TSO, the application would typically output a message like: "Allocating datasets, please be patient". That is because executing a statement like "ALLOC DD(xxxx) DSN(xxx.yyy.zzz)" took a half minute to a full minute to execute. The MSS monstrosity was amazing to watch in operation. It consisted of a wall of honeycomb-like compartments where the tapes were stored. The operation of the whole contraption was mostly transparent to the user. The command to allocate a dataset would cause the picker to locate the honeycomb location where the tape was stored, retrieve it, and mount it on a device that emulated a DASD. I don't think the tape was copied to a real DASD. Data was read from and written directly to the tape, which was quite wide. I think each vertical (crosswise) stripe on the tape emulated a DASD track. When a FREE command was executed, the tape would be rewound back into the cartridge and the picker would put it back into the honeycomb. I'm not sure if any special support would be required from Hercules. I think the VOLSER of the tape appeared to the OS as just another 3330 DASD. (This is all from my distant memory of conversations with older engineers would used to work on the beast. I'm sure an internet search would yield a much better description.) Charles Bailey On 2025-02-07 00:21, Dave Trainor - N8ZFM wrote: So I am studying the OS JCL and Utilities book (Michael Trombetta) from the mid 80's, good book ,was recommended to me for the MVS 3.8j era of JCL. Am happy with it and can also recommend it.
They discuss utilization of the MSS storage device(s) a 3850, and at the mid 80's of the book it was a current device, seems to consist of a 3330 disk (maybe more than one) and a series of tape carts, with an "invisible" autoloader system, so as to implement "near-line" storage. Only one actual dasd and a bunch of different images of it on tape, that would be recalled by the VOLSER from MVS and loaded on the live DASD where it was then a live 3330 disk. (obviously a delay in preforming this "staging operation" as I take it was called).
Probably more complicated in reality but that is my gist of the functionality. Different logically but easier than mounting a tape, doing a restore, then archiving, not as fast as having dedicated DASD.
I assume since I can't find it in Hercules, that this device and not sure to call it tape or disk - is not emulated by Hercules, probably would not be useful and no one uses any such scheme since the mid 80 As the need probably got replaced quickly by vastly larger amounts of cheap DASD being available? Anyone know? Or remember these things and why they are obscure today?
I do not see this as being the same functionality as a backup tape library (automounter) of true tapes treated as tapes which is for sure still being utilized.
Is my assessment of what this 3850 MSS device was correct? And why we don't have an emulation for such a thing anymore?
Thanks,
Dave N8ZFM
|
Jeff Bassett
Frank R
