2022 - Running a mainframe on your laptop for fun and profit

Yes, this talk is about running your own mainframe on your own hardware. Mainframes are old, yes, but they are still very much alive. New hardware is still being developed and there are a lot of fresh jobs in this area too. A lot of mainframes run COBOL workloads. COBOL is far from a dead language. It processes an estimated 85% of all business transactions, and 5 billion lines of new COBOL code are written every year. In this session the speaker will help you in take your first steps towards running your own mainframe. If you like then after this session you can continue to build your knowledge of mainframe systems using the links provided during the talk. Come on in and learn the basics of a completely different computer system! And it will take you less than an hour to do that!

Yes, this talk is about running your own mainframe on your own hardware. Mainframes are old, yes, but they are still very much alive. New hardware is still being developed and there are a lot of fresh jobs in this area too. A lot of mainframes run COBOL workloads. COBOL is far from a dead language. It processes an estimated 85% of all business transactions, and 5 billion lines of new COBOL code are written every year. In this session the speaker will help you in take your first steps towards running your own mainframe. If you like then after this session you can continue to build your knowledge of mainframe systems using the links provided during the talk. Come on in and learn the basics of a completely different computer system! And it will take you less than an hour to do that!

Jeroen Baten

#mch2022 #MCH2022Curatedcontent

[reformationfan]

Great talk, I am a retired mainframe software engineer. The operating system you demonstrated on is MVS 3.8 which was current in the mid 80's, in fact the put level of the software was 8505. This might have been the last release of MVS before MVS/XA which you mentioned which extended the address spaces from 16M to 2G. The latest releases of zOS (last time a checked) do run on the Hercules emulator which is no toy. I know of a company (over 20 years ago) which used Hercules for a few months before they could acquire a Z800 mainframe.

[garanceadrosehn]

There's also the "Michigan Terminal System" (MTS) which was developed by a consortium of Universities. It multuser timesharing system which was *mostly* open-source and developed and in production into the mid-1990's. It could support a few hundred users at the same time, given a big enough mainframe. I know some of my friends are still running it on top of Hercules.
Thanks for the talk. I hope to come back to check out Hercules sometime soon!

[SuperHaunts]

Thanks SO MUCH for reminding me just how far we've come with computers & user-friendliness! How I functioned back then, I can't believe.

[bernardvanderhelm]

Hi Jeroen, great speech! Great to see that hercules-390 is still in use!

[sundhaug]

Fujitsu still sells mainframes, but have announced end-dates for sales and support

[oneeyedphotographer]

Dave Plummer (Daves Garage) has a video where he visits IBM's manufacturing plant. I highly recommend it.

[joev]

DASD = Direct Access Storage Device, not "attached".

[alexloktionoff]

From my point of view mainframes are transactions crunchers, simple throughput number corners are supercomputers.

[sturm]

Great talk - thankyou. Mainframes are alive and well. Banking, automotive design & manufacturing, any large business with large databases - they all have mainframes. And the pay is much higher than for PC programmers because of scarcity.

[maschwab]

MVS380, VSE380, and VM380 is available, partitions V=R memory between the 16M and 2G line.
TCP/IP is scheduled to be included in TK5 Update2 by Jan 2024.
Intercomm (cics clone) is now included.

[kevincozens]

I ran some Fortran programs on an IBM 360 and 370 back in the day. I also did a little bit of programming in assembler for the machines. I still have the book on IBM 360 Assembler language. It could be fun seeing if Hercules could be used to run some old programs.

[bobdobalina]

"You stick that into a token ring hub, which you have laying around, as one does." LMAO.

[oneeyedphotographer]

Datasets are files, and users create their own more often that using those created by others, system datasets excepted.

DASD aren't necessarily hard disks, emulated or real. There used to be drums, real and emulated.

[smakfu]

The mainframe is great for many reasons, but why do its cheerleaders always need to do phallus sizing comparisons, especially when there's a lot of fancy hardware beyond IBM's moat. A Z16 Telum four drawer systems is (physically) 32 compute processors, with 256 cores and 512 threads with a max of 40TB of RAM. Which is fine, especially since their IPC throughput is impressive and raw clocks are high, but if brute compute bragging rights is what you're about, then commodity x86 (in specialized form) can easily beat mainframe hardware. You can get a shared-memory modular NUMA chassis Bullsequana-SH320 that will scale up to 32 sockets (2 sockets by 16 modules connected via 2x UPI at 11.2 GTs, or 166 GBs) at 60 Sapphire Rapids cores per socket for a total 1920 cache coherent Golden Cove cores and 3840 threads (logical processors) with 128TB of RAM (all accessible from a single OS instance), which dwarfs the Z16. And again, this is in a single shared memory system that can also support 32 GPUs. Love or hate Intel, but Golden Cove cores are exceedingly powerful, comparable with Telum IPC at 6 decode, 10 issue, versus Golden Cove at 6 decode, 12 issue. A 32 socket, 1920 core Sapphire Rapids system is one hell of a monster.

However, Telum has an obvious advantage in clock speed and a highly sophisticated inter-processor shared caching architecture. Telum also features interesting shared on-die inference accelerators that are especially well suited for certain types of overlay transaction analysis, actions that would be much harder to implement (from a code perspective) on Golden Cove cores, using per-core SIMD. But the REAL reason you buy a mainframe is for unmatched hardware and software integrated fault tolerance, resilience, IO interconnect and partitioning capabilities. While the commodity world has virtualization and containerization, it doesn't have anything nearly as sophisticated or configurable as LPAR's. With firmware support, the ability to carve out and configure resources for LPAR use (with firmware level dispatching and specialized support processors) is unmatched. Furthermore, this is all backed by in-frame processor level fault recovery with processor spares, ICF's for sysplex fault tolerance, etc., which simply doesn't exist outside the mainframe world. And it's all tightly coupled in a very expensive, vertically integrated stack of hardware and software. You pay for a mainframe because, as the old saying goes, nobody ever got shot for buying IBM. The mainframe world doesn't talk about four versus five nines, they talk in terms of cash reimbursements from IBM if ADP's payroll operations are impacted by a failure.

So no, I don't think anyone should be talking about the mainframe because it's the biggest or baddest of computers (the most powerful shared memory computers are not mainframes). The mainframe is impressive because it's designed to run 24/7 and never, ever, fail in its intended purpose.

[mrpugster]

holy shit, I've not seen those screens since 1996 haha

[Drew-Dastardly]

Can I suggest simh as an excellent mainframe emulator? I have a DEC-10 and a PDP-8 permanently running as background jobs. I telnet in from time to time.

[augurcybernaut]

Guess need to breakout my old CICS Copy Books

[TheRealNewBlackMusic]

Does it have an option for CICS and vsam db2

[rndofpipowe]

Where is the disco-ball???
That was my first thought 😊

[frankniethardt]

Actual TSO and CICS are different frontends running on top of OS/390.