We've come a long way

For fun this evening, I was reading about how to install a mainframe computer at home. Obviously, not physically: Real mainframes (the ones made by IBM, Hitachi, Siemens/Fujitsu, Amdahl, CDC...) are physically so large and use so much energy, they are impractical for a normal house. I used to do a lot of work on an IBM 3084, and not even the CPU box (without storage, memory, tapes, or networking) would fit into our living room. Matter-of-fact, the water needs of that computer (it was obviously water cooled) would leave us unable to shower or wash our hands, and purchasing one in the early 1980s cost about $10M. Instead, today one can install the Hercules emulator, which runs the IBM mainframe instruction set on any reasonable computer, and is fast enough.

That then leaves the problem of an operating system. Fortunately, IBM has released the most-commonly used operating system MVS into the public domain. That includes important facilities such as
  • JES2 (which is needed to run any jobs on the OS, and without the ability to run a program a computer is utterly useless),
  • TSO (which allows interactive use from a terminal, otherwise the only way to use the computer is to submit (emulated) card decks to start batch jobs), and
  • compilers (for things such as Assembly, RPG-II, COBOL, Fortran, Pascal, Snobol, and Spitbol, no I am not kidding).
  • Unfortunately, the scripting language REXX does not seem to be available freely, so scripts have to be written in CLIST, which is painful.
Then I looked how to set up and install the system. This is much more involved than on modern computers. Fortunately, I found a set of instructions in roughly 10 very long web pages, written by Jay Moseley. I think it would take me several weekends to get through them (it has been 20 years since I last used a mainframe, and 30 since I helped administer one), but the level of detail he gives is amazing. After reading all these pages, he shows how to enter a "hello world" program in COBOL (using an interactive editor that is part of TSO):
mvsCap31.gif


Then one needs a batch job to actually compile, link and run this program (this is the equivalent of the "./a.out" command that one would enter into a Unix system to run the program just entered). Yes, you have to type in everything that's in green below:
mvsCap36.gif


And then, finally a few more TSO commands, and one can view the output of the program. And verily, it says "HELLO WORLD!" in all upper case.

About 35 years ago, I thought being able to compile and run a program was a big deal, and very productive. Today, looking at these instructions, and the amount of work required to just print "hello world", I can only laugh. We have come a long way.
 
For fun this evening, I was reading about how to install a mainframe computer at home. Obviously, not physically: Real mainframes (the ones made by IBM, Hitachi, Siemens/Fujitsu, Amdahl, CDC...) are physically so large and use so much energy, they are impractical for a normal house.

Unless you bought one of these :).
 
Even in the 1980s, there were small air-cooled mainframes that would fit in a household. The IBM 43xx series didn't need a raised floor, and Honeywell had a GCOS machine that was even smaller. I had forgotten about the 390-on-a-card that Balanga mentioned; but I know about the 370 that fit on three PC-bus cards, using multiple modified 68000 chips (our department had one of those).

A friend of mine manages a group of volunteers at the Computer History Museum, where they have restored an IBM 1401; it was the most popular IBM computer before the 360, built in 1959..., and thousands of them used to exist. With the help of dozens of volunteers, and by finding two complete machines that were in operation into the 90s, they managed to create two working machines.
But as Crivens points out, software and documentation is much more difficult.
 
Interesting subject matter; and brings back memories of my experiences with IBM. Other than a timeshare teletype modem from math class and elsewhere; IBM was probably the first computer operating system I worked with.

EDIT: And it is a reason why I still like the Thinkpad laptops.
 
It's amazing to watch the 1401 compile a program. Requires 10 minutes of pressing buttons. Like rewinding the compiler (on the tape drive) before using it a second time, putting the source code in the card reader, and making the printer ready before the program runs. This makes a 1970's mainframe feel positively convenient.
 
I don't know. Even looking at that cobol source makes me somehow wanting to wash my eyes. We came a long way indeed.
 
Interesting subject matter; and brings back memories of my experiences with IBM. Other than a timeshare teletype modem from math class and elsewhere; IBM was probably the first computer operating system I worked with.

IBM 'Operating system'?

EDX, DOS/VSE, MVS, OS/400 PC-DOS, OS/2, AIX, z/OS?

There's quite a lot to choose from ;)
 
I still think about how I laughed at the friend of mine who, in the late 1970s, went to community college to study COBOL and, till he died last year, was still programming in it for the financial institution he worked for.
 
I'm betting he made lots of money the last couple of decades. I was already working quite a few years in IT when that Y2K stuff became important. The insurance company I worked for at that time had a mainframe with a LOT of COBOL code running on it. They had to ask a couple of pensioners to come in and fix the code because they were the only ones that knew COBOL. Same thing happened a few years later with the Euro-conversion. Needless to say, these old guys were paid huge amounts of money.
 
balanga, Yes, I had the pleasure (and pains) to work with DOS/VS on IBM 360/370.
Pretty simple machine and parts inventory control systems (MAPICS) that were useful for auto and trucks parts warehouses. I know I had two three-ring binders of the stuff, but it is packed away in storage bins right now.
 
Back
Top