the good old days of mainframe computers

[Don Shepherd]

This is from the book "A Few Good Men from Univac". I think many of us old-timers can identify with this.

The typical customer for a large computer was and still is a large scientific or engineering firm, a government laboratory, or a university. The large computer system would be installed in a climate-controlled computer room with a raised false floor to accommodate the interconnecting cables. The computer center was isolated from the remainder of the building and staffed by a variety of computer specialists. It was often surrounded by glass panels, thus being visible but tantalizingly inaccessible to the average engineer or scientist. The only way that a typical user could get a job run was to physically carry a deck of punched cards or reel of magnetic tape to a counter at the computer center. There an attendant would take the job together with such information as name, account number, interplant mail station, and phone number. Depending on the computer center workload that day and the user's priority and/or political influence within the organization, the job would be inserted into the card reader attached to the computer and either run immediately or stored temporarily in a queue of jobs on magnetic drums or disks. Nearly all the computer centers of the 1960's were run "closed shop"; that is, only the computer operator staff was allowed to enter jobs into the computer and retrieve the results, most usually a stack of printouts from the high-speed printer. The user's results would eventually be printed and returned by interplant mail or by a special courier service run by the computer center, but the elapsed time between submitting a job and receiving the results could often be two or three days. Users complained bitterly about this system with comments such as "What is the use of running my job on a computer that can solve the problem in thirty seconds if it takes me three days to get my results? I can run the job on a desk calculator in three days!"

Those were the days, my friend ....

[Albert Chan]

The book is available from the Computer Museum

http://tcm.computerhistory.org/exhibits/FewGoodMen.pdf

[Maximilian Hohmann]

Hello!

(09-21-2021 04:52 PM)Albert Chan Wrote:  The book is available from the Computer Museum...

I downloaded the book but I don't think that I will ever read it. Wasted enough lifetime already with those mainframes themselves...
At university we were allowed to use card punch machines and punchcard readers by our own. But I remember an internship that I did in the aerospace industry 1982/83 where normal users were not even allowed to punch their their own cards. One had to fill-in 80 column paper forms in writing and typists would then operate the card punch. Before sumbitting the job they handed back the stack of cards to check for typos and if everything was OK it got submitted. When I came back to university from the internship I found to my pleasant surprise that the card punches and readers had been replaced by "time-sharing" terminals.

Regards
Max

[Ren]

Thanks for the link!
(downloaded)

[Bill (Smithville NJ)]

Don,

It's a great book. i have read it several times.

Another great book on early days of Univac is "From dits to bits : a personal history of the electronic computer" by Herman Lukoff. Tells his personal story on being an engineer designing, repairing and maintaining the Univac computers. You can check out a copy from the Internet Archive Library:

Borrow From Dits To Bits

73
Bill WD9EQD
Smithville, NJ

[Don Shepherd]

I well remember working on such a mainframe Univac large computer at my first job. Rumor had it that there was a systems guy who crashed the computer on purpose (and it took much more than a divide by zero in a user program to do that). Upper management wanted to fire him, but the systems people knew how valuable he was and they overruled upper management.

[Ren]

(09-22-2021 02:09 AM)Don Shepherd Wrote:  I well remember working on such a mainframe Univac large computer at my first job. Rumor had it that there was a systems guy who crashed the computer on purpose (and it took much more than a divide by zero in a user program to do that). Upper management wanted to fire him, but the systems people knew how valuable he was and they overruled upper management.

At a former employer, one of my cow-orkers found a program called "CrashMe" on the Internet. (This was pre-WWW). And decided to try it on his Sun workstation.
Unfortunately, he cut and pasted it into a window that was logged into our main Sun server!
(No real harm done, other than a bunch of upset cow-orkers!)

[Dave Britten]

I always find mainframes interesting, both in and of themselves, and also as adjacent to the era of early programmable calculators (the first "personal computers") democratizing computing and allowing a scientist, engineer, financier, etc. to produce, test, and refine results from an HP 25 or SR-52 or what have you in just a matter of hours, rather than sending a card stack to the computing center and hoping to get usable output the following day.

I don't have the space - and my wife doesn't have the patience - to ever acquire old mainframe hardware for our basement, but perhaps someday I'll pick up something in the minicomputer category. An old PDP or AS/400 would be neat to have and tinker with.

[Bill (Smithville NJ)]

(09-22-2021 02:09 AM)Don Shepherd Wrote:  Rumor had it that there was a systems guy who crashed the computer on purpose

I remember our first Sun Unix server at work. We had an outside firm that maintained it. They had set up a back-up system (to tape at that time) that I couldn't understand how it worked or even verify that it actually worked.

So one day I informed them that next week i planed to deliberately hard crash the system and then we would verify that they could do a complete restore from their crazy back-up system. They were horrified and finally admitted that their system did not do a full system backup.

A co-worker and I then proceeded to spend a very long weekend reading the Unix manuals and learning the correct way to do a full system backup. We then replaced their backup with the correct backup and fired them.

And yes - I did do a hard crash of the Unix System and then used the the backup to restore it. Worked perfectly.


One other side note. When I was in college in the early Seventies, there was an unofficial "Crash Club". To become a member, you had to crash the large student CDC computer to the point that it had to be powered off and then brought back on-line. I was a proud member of the Crash Club.

The interesting thing was that you really didn't get into bad trouble for doing this. In my case, I ended up doing an "Independent Study Course" with a professor where I studied why my method crashed it and how the system could be fixed so that way would not work in the future. I even got course credit towards my degree.


73
Bill WD9EQD
Smithville, NJ

[bbergman]

(09-21-2021 05:21 PM)Maximilian Hohmann Wrote:  At university we were allowed to use card punch machines and punchcard readers by our own. But I remember an internship that I did in the aerospace industry 1982/83 where normal users were not even allowed to punch their their own cards. One had to fill-in 80 column paper forms in writing and typists would then operate the card punch. Before sumbitting the job they handed back the stack of cards to check for typos and if everything was OK it got submitted.

Mainframe life was very different, but I do think it (appropriately) contributed to the maturity of software development over time. Yes, it was a closed system, but that was mainly an asset protection factor. It wasn't like you could buy a new laptop for $500; this was a piece of equipment that cost $50k-$100k per CPU. You had to protect it.

When I was learning Fortran in high school, we punched the cards in class and sent the stack out for processing. 3-4 days later, we got our cards back, plus a green+white 132 column printout from the line printer that showed our run results. It only took ONE buggy run to teach us the value in hand checking our code before sending it out. Within a couple of weeks of the start of the semester, everyone was going over their code, running it on "paper", checking for boundary conditions, doing unit testing, pair programming, and more. No one wanted to send in a stack of cards and have them come back next week with a failure on Line 1. These practices all translated into a maturing software development mindset that is still evidenced today. Maybe not all the same practices, but many of them are. It was an important step.

[Don Shepherd]

(09-22-2021 05:11 PM)bbergman Wrote:  Mainframe life was very different, but I do think it (appropriately) contributed to the maturity of software development over time. Yes, it was a closed system, but that was mainly an asset protection factor. It wasn't like you could buy a new laptop for $500; this was a piece of equipment that cost $50k-$100k per CPU. You had to protect it.

When I was learning Fortran in high school, we punched the cards in class and sent the stack out for processing. 3-4 days later, we got our cards back, plus a green+white 132 column printout from the line printer that showed our run results. It only took ONE buggy run to teach us the value in hand checking our code before sending it out. Within a couple of weeks of the start of the semester, everyone was going over their code, running it on "paper", checking for boundary conditions, doing unit testing, pair programming, and more. No one wanted to send in a stack of cards and have them come back next week with a failure on Line 1. These practices all translated into a maturing software development mindset that is still evidenced today. Maybe not all the same practices, but many of them are. It was an important step.

Bruce, we called that "desk checking" and our boss insisted that we do that. Yes, it took some time, but it invariably caught coding and logic errors that might not have been detected otherwise. Our US Census Bureau FORTRAN programs processed files with millions of records (one record per business establishment in the US) and reruns due to program errors would have been unacceptable.

[robve]

Ah, the "good old days". Brings back memories when a simple forkbomb brought the entire system down. Some students in my class in the 80s did that to annoy other students or to get a project submission deadline moved due to complaints that the system was down again. Because we had dumb terminals, some not-so-ethical students stole passwords (and project source code consequently) with a simple Trojan horse shell script they had written that looked exactly like the legit green ASCII login screen. Others were manning the shared printer to get their hands on someone's project printout. Account quotas were ridiculously small (I don't remember for sure, but I think it was a couple of kilobytes). Computing was a lot more "physical" back then in many ways.

- Rob

[BobVA]

(09-22-2021 09:02 PM)robve Wrote:  Ah, the "good old days". Brings back memories when a simple forkbomb brought the entire system down....

Nothing like hearing the "click" of all the 2741 Selectric terminals locking up to send a shot of adrenaline through the heart of a student working on a late assignment. :-)