Reminiscing about the mainframe days ...

vermaden said:
Most of today's operating systems are mostly useless without a proper Internet connection,

Yeah.. ultimate regression or what? lol

Kinda like going back to the single mainframe connected to by lots of little terminals. The only difference now is that we use cat5 rather than rs-232.
Then again, UNIX has always been like this.. I guess it is only Windows with it's online activation and Mac OS X with it's dumb app-store that have regressed.

But I digress..
[/offtopic]
 
Yeah but mainframes never used RS232. Anyway we used BNC connectors and coax for terminals in the olden days.
 
randux said:
Yeah but mainframes never used RS232. Anyway we used BNC connectors and coax for terminals in the olden days.

There was a little company called DEC that used serial terminals.
 
Yeah but that little company made minis, not mainframes. The biggest machine they had was about the size of an IBM 1403 line printer or a small stack of 9 track tape drives.
 
randux said:
Yeah but mainframes never used RS232. Anyway we used BNC connectors and coax for terminals in the olden days.

Randux, back in the 70's we used the PBX's 66block and twisted pair for IBM terminals that were connected to our IBM UNIX mainframe using SNA as the transport protocol. Oh the days of PU's and LU's and let's not forget NetGen's. Yeah, I am old fart!

Just-Mike, as far as your list of requests, please remember that FreeBSD has been around for almost 18 years. Some people take offense to some of your requests because we've done UNIX for a long long time and FreeBSD is as close as you can get to True UNIX with out paying for it. But, I do agree with you on improving the Handbook with more photos and graphics. Please join the Documentation team and send PR's. That would be a good start.
 
rbelk said:
Randux, back in the 70's we used the PBX's 66block and twisted pair for IBM terminals that were connected to our IBM UNIX mainframe using SNA as the transport protocol. Oh the days of PU's and LU's and let's not forget NetGen's. Yeah, I am old fart!

Hi rbelk. As far as I know, UNIX never ran on any IBM mainframe. AIX, which is IBM's UNIX, didn't come out until 1986 and even then, it ran only on their minis, like S/34 and S/36. As far as I know, token ring (twisted pair) didn't come out until about the same time and I personally never saw it deployed at all with 3270s, it was only when we started using PCs as dumb terminals that people used twisted pair. Before that it was all coax.

So exactly what OS and hardware are you talking about?

I started coding on OS/MFT on a 370/168 among other classic iron, so if you're an old fart then I am as well :p
 
randux said:
Hi rbelk. As far as I know, UNIX never ran on any IBM mainframe. AIX, which is IBM's UNIX, didn't come out until 1986 and even then, it ran only on their minis, like S/34 and S/36. As far as I know, token ring (twisted pair) didn't come out until about the same time and I personally never saw it deployed at all with 3270s, it was only when we started using PCs as dumb terminals that people used twisted pair. Before that it was all coax.

So exactly what OS and hardware are you talking about?

I started coding on OS/MFT on a 370/168 among other classic iron, so if you're an old fart then I am as well :p

I started as an operator on an IBM 370/155 and 370/158. I then went to application programming and from there system programming. Yes, I have used TSO, CICS, DB2, Cobol, jcl, REXX, Panvalet, EasyTreive, and 370 assembler. I also did RPG on an AS/400 and System 36 and 38. I stay confused!

It was also more economical to use the phone system wiring than to buy the coax back then. All you had to get a coax-to-rj11 adapter and wire it through the phone wiring that already existed.

I'm going to have nightmare's tonight about twin-ax and punch cards, aaahhhh!!

Yes you are right, it was in the early 80's and we bought an Amdahl 580 and it ran UTS in an LPAR. UTS did come out in the 70's but it wasn't commercial until '81 or '82.
 
rbelk said:
I started as an operator on an IBM 370/155 and 370/158. I then went to application programming and from there system programming. Yes, I have used TSO, CICS, DB2, Cobol, jcl, REXX, Panvalet, EasyTreive, and 370 assembler. I also did RPG on an AS/400 and System 36 and 38. I stay confused!

It was also more economical to use the phone system wiring than to buy the coax back then. All you had to get a coax-to-rj11 adapter and wire it through the phone wiring that already existed.

I'm going to have nightmare's tonight about twin-ax and punch cards, aaahhhh!!

Yes you are right, it was in the early 80's and we bought an Amdahl 580 and it ran UTS in an LPAR. UTS did come out in the 70's but it wasn't commercial until '81 or '82.

Nice to meet you, there are still plenty of us around. I am still writing systems software in assembler on IBM systems, same job since the 1970s! Things have changed a lot but in a way they haven't.

I never worked on an Amdahl, I know Gene was supposed to be a brilliant guy. But I have worked only on IBM machines all these years except for some porting I did to Fujitsu. Now I am trying to learn Sparc. I guess I only like dinosaurs and stuff that not too many other people care about or work on!
 
Randux, I bet you still have a Z-80/CPM system, haha. Wait, what am I saying, mines in the top of the closet...

Sorry Mod's but I had to divert a little:p
 
No, but I wish I did. I finally got interested in the old microprocessor systems since I never used them in their heydey.

Maybe we should start a project to port FreeBSD to z80. Now we are back on topic even for this thread!
 
randux said:
Maybe we should start a project to port FreeBSD to z80.

You mean I no longer need to spoof?

% host -t hinfo spectrum.skysmurf.nl
Code:
spectrum.skysmurf.nl host information "ZX Spectrum" "Sinclair BASIC"

Fonz
 
I'm partial to IBM iSeries (f.k.a. As400 et al). I didn't do any punch cards, but I do remember the coax/twinax dumb terminals.

"Mainframe", these days, might apply more specifically to the hardware than the OS running it. And even still the Power hardware line scales out into super-computing territory..
 
I remember when the IBM PC was starting to be used by business. The main selling point was you didn't have to use a connection to the mainframe! Now we have PC's that depend on connections to multiple servers and a heftier network to support them. Decentralization is a bad word again, and Information Systems executives say we need to be under their control again.

The VMware VSphere system is a good example of this. Heck, we did the same thing with LPAR's and WPAR's back in the early 80's.

Those who cannot remember the past are condemned to repeat it, and Microsoft is leading the way...
 
fonz said:
You mean I no longer need to spoof?

% host -t hinfo spectrum.skysmurf.nl
Code:
spectrum.skysmurf.nl host information "ZX Spectrum" "Sinclair BASIC"

Fonz

LOL Fonz, good one! :e
 
qsecofr said:
I'm partial to IBM iSeries (f.k.a. As400 et al). I didn't do any punch cards, but I do remember the coax/twinax dumb terminals.

"Mainframe", these days, might apply more specifically to the hardware than the OS running it. And even still the Power hardware line scales out into super-computing territory..

No, mainframes are still around and still run the great grandson of the same basic mainframe OS they did from the 1960s. Of course because of the money involved there have been many updates and technologies added. Any good definition has to specifically relate to the hardware and software together because they're tightly coupled. Mainframes aren't supercomputers because nobody has made the effort to cluster them. But on a single box, no system I know of is more secure or can get more general computing work done than a mainframe. And they're fantastic systems for developers, the quality of the software and tools is outstanding. I still work on them today and although there are not as many being sold as there were twenty years ago, they are sold and so is new software and hardware.

--
Mainframes: Yesterday, Today, and Tomorrow
Quality, Performance, Availability, and Security for the next century
 
randux said:
Mainframes aren't supercomputers because nobody has made the effort to cluster them.
Your sentence is unclear thus can be misleading.
  1. Most of the time "supercomputers" and clustering is the same as number crunching (HPC). In that regard, mainframes aren't supercomputers just simply because their CPUs can't compete with simple Intel-compatible hardware when it comes to cost per "number-crunch" capacity. In short; if you want number crunching capacity; buy a cluster of Intel-compatible machines.
    What mainframes are really good at, is I/O capacity; a mainframe can "push" much more data in and out than any Intel-based machine can.
  2. of course mainframes can be clustered, for IBM mainframes the capability dates back to 1994, look at Parallel Sysplex and GDPS.
 
rbelk said:
and Information Systems executives say we need to be under their control again.
...
Those who cannot remember the past are condemned to repeat it, and Microsoft is leading the way...

It is odd but people seem to love being under control of severs belonging to large companies so that they can just sit there complaining like a bunch of morons when it goes wrong.

A good example is when the 4chan ddos skript kiddies clogged up playstation and xbox servers and people couldn't log in to use their purchased stuff. People just started whining rather than questioning why they needed to be under control of such central servers to play video games.

After all, my SEGA megadrive works great offline! :)

(I also kinda bundle linux central package repositories in with this insanity, which is why I love the ports collection fetching the source from the distributors servers rather than a central server)
 
tingo said:
Your sentence is unclear thus can be misleading.

Yeah I guess it could be unclear if you get all your mainframe info from wikipedia. From your post I would say you just like arguing about stuff you have no personal knowledge of (meme) and think a box based on a 50 dollar chip can outperform a box that costs half a million dollars for an entry level system. I think anybody who believes that needs a serious make-over :p Get a clue, then come back and see us!
 
randux said:
No, but I wish I did. I finally got interested in the old microprocessor systems since I never used them in their heydey.

Maybe we should start a project to port FreeBSD to z80. Now we are back on topic even for this thread!

Randux, there is a UNIX work-a-like that runs on the z-80! It's called UZI. I have never used it, but I've always wanted to.
 
OK Guys, I found this statement on Wikipedia's Unix-like web page.

  • IBM z/OS's POSIX compatibility layer is sufficiently complete to be certified as trademark UNIX

I know it came from Wikipedia, and I don't trust Wikipedia. Does anyone know if this is a valid statement? Chime in anytime randux!

Since this is an off topic thread anyway, I'll post my links to old and odd UNIX like systems.

Dave Braun at Intel wrote UZI (a unix v7 ish system for Z80 processors, and public domain) - one of the gnu project mistakes was not building off this but using Mach
http://www.dougbraun.com/uzi.html

Steve Hosgood at UW Swansea wrote OMU - a 6809 and later 68000 based platform with a unixlike api
http://tallyho.bc.nu/~steve/

Jawaid Bazyar of EGO systems wrote GNO/ME a unix like Multitasking Environment for the APPLE IIgs 65816 computers.
http://www.gno.org

commodore/64 - Daniel Dallmann, lunix
http://lng.sourceforge.net/

msx - Adriano C. R. da Cunha, uzix
http://uzix.sourceforge.net/
 
Hi rbelk. Yes, USS (Unix System Services) came out I believe sometime in the 1990s. They seem to be trying very hard to rename it because of a conflict with very old VTAM terminology but even the current doc still calls it by the same name. It is part of z/OS. AFAIK it is implemented as a subsystem and may have a few address spaces associated with it. You can not run it standalone, you have to license and run z/OS.

I believe they did this is to try and preempt people from abandoning the mainframe just because they wanted to run UNIX. It wasn't successful in that (actually they did gain bare-metal or virtualized Linux workloads eventually on mainframes but that doesn't use the UNIX subsystem!) but what it does allow you to do is some interesting interoperability between UNIX and z/OS, since z/OS apps can make UNIX calls and the UNIX apps can share *some* z/OS filesystems and other stuff. Personally I believe licensing issues were/are part of the issue since IBM's version is licensed and they really don't care to mix open source code with their code. It's much easier to run somebody else's Linux on their hardware than it is to use somebody else's compiler/app/what have you on *their* OS running on their hardware.

It turns out people are hotter on Linux than UNIX on the mainframe, so I'm not sure how much longer IBM will continue to support USS or how much work goes through it. I do know we haven't been asked to do anything with it ever, indicating big sites are not using it very much.

If you don't stop talking about UNIX you are going to go back on topic and then somebody will have to move the thread again!
 
randux said:
Yeah I guess it could be unclear if you get all your mainframe info from wikipedia. From your post I would say you just like arguing about stuff you have no personal knowledge of (meme) and think a box based on a 50 dollar chip can outperform a box that costs half a million dollars for an entry level system. I think anybody who believes that needs a serious make-over :p Get a clue, then come back and see us!
I used Wikipedia as a reference because it was handy. I didn't claim it was my only source of information. I also haven't said anything about my experience with mainframes. :)
I'm very happy to see some facts and other references to back up your claims, if you have them. :)
Hint: anybody who thinks that the most expensive system gives the most performance in any situation is in serious need of a reality check. :)
 
My first programming course was FORTRAN Programming using an IBM mainframe (a very small mainframe, 43XX series if I remember correctly), and you guys were very lucky if you had a terminal with a serial or a coaxial cable. I used a punch card machine to program and I had no direct connection to the mainframe. I submitted the deck of punch cards to the operator and the result (printout from a line printer) was back in about an hour later. From junior year, my school got enough IBM terminals and no more punch cards necessary. I still have punch cards from those days. They are very good bookmarks.
 
randux said:
Yeah but that little company made minis, not mainframes. The biggest machine they had was about the size of an IBM 1403 line printer or a small stack of 9 track tape drives.

Um, no. DEC's biggest machines were the PDP-10 mainframes, which were comparable to a mid-range IBM mainframe. DEC cancelled new developments of their mainframe line in 1983 in order to put more resources into the Vax, but it was 10 years before the biggest Vax was as powerful as the largest PDP-10 (and the Vax always remained a mini).
 
No, you're totally lost. Nobody at DEC ever referred to anything they made as a mainframe. Do you really want to argue with Ken Olsen? ALL the PDP machines were minis and none of them ever competed with mainframes. I've used DEC 10s and 20s. They were not mainframes. I understand how a guy who gets all his "information" from WikiPedia and never used a DECSystem 10 or an IBM mainframe could confuse them. But they really have nothing in common except they're computers.

https://research.microsoft.com/en-us/um/people/gbell/Digital/DECMuseum.htm
http://hightechhistory.com/2011/02/08/ken-olsen-co-founder-of-digital-equipment-corporation-dies/
https://en.wikipedia.org/wiki/Digital_Equipment_Corporation
http://www.cwhonors.org/search/oral_history_archive/ken_olsen/index.asp
http://www.computerhistory.org/revolution/minicomputers/11/335
http://sloanreview.mit.edu/improvis...ns-from-ken-olsen-and-digital-equipment-corp/
 
Back
Top