IBM Mainframe Forum

[alanplayford]

Just interested, folks .....
Any of you work with or for a mainframe software development company who DO NOT have their own mainframe, alebit small, or a zPDT, or their own access to an LPAR?
If so, what difficulties have you experienced in testing, roll-out etc?

[jackfarrah1991]

No, I have never encountered such a thing...

[steve-myers]

zPDT appears to most folks as similar in over all goals to Hercules, which is why IBM foams at the mouth at any mention of Hercules as Hercules runs z/OS quite well, though with serious limitations on the number of concurrent users it supports relative to z/OS on a real Z machine. JES2 and PSF2 development requires real hardware because you can't attach real 3800 / 3900 or modern 42xx printer class machines to a PC. Similarly hardware support for new tape and disk devices requires a real Z machine. But all of that is IBM.

Back when I was CA, most of us used an LPAR or zVM based virtual machine out on Long Island, but we didn't need hardware product code. I have no idea what the tape folks use for their product testing. I do recall an interview with the CA1 boss inquiring about our product's use of tape. Long after the interview I realized this was related to the CA virtual tape product; I recall he seemed happy after we assured him we used straight QSAM, BSAM and standard JFCB modifications for tape positioning; nothing exotic.

Not too long after I started with CA there was talk about moving off the zVM platform to Hercules, but that apparently went over like a lead balloon; our products were very dependent on sharing a few pseudo 3390s between the two LPARs dedicated to product testing and the zVM systems we used for more ESA390 and zOS level testing. I didn't realize how dependent we were on central support to prepare the ESA390 and zOS SYSRES volumes when new versions of the system were released to us ISVs when this discussion started at the time; with Hercules this support would have ended.

This did not always work well. I recall one zOS release where JES2 control block offsets we used changed between the first beta and the final beta and we didn't catch it; we had to hurry up and fix our PTF for that zOS release. In z/OS 1.7 another minor product didn't work. I didn't know we supported that product and I hadn't tested the damn thing. We didn't know about it until the ACF2 folks called up and told us our product - source released as it was all JES2 exits - messed up ACF2. Well, after some belated research, we determined it had to be rewritten from scratch. One of my colleagues and I worked up two independent solutions. He got there a day or two before me; my solution was easier to install into JES2 and was a little more elegant in other respects, but Joe got there first. Sigh. The difference, I think, is Joe had never worked with installing modifications in JES2 or JES2 exits and I had.

[steve-myers]

Speaking of real printers, does IBM still support 3800-1 (the very first 3800)? 3800-3 which ushered in PSF2 and FSS. I quickly realized that the whole goal was to get new printer support away from the JES products. I do know that 3800 support altered HASPRPU in the 1970s in ways that us old timers never quite got used to and PSF2 style support in JES2 - well, I'd rather not think about that!

I think the 42xx printers go through PSF2 even though the output looks like a line printer to me. But I only saw one back in the 1990s when I worked for Sungard. They are wicked fast compared to the 1403 model 3 (or -N1). You really needed ear protection near a 1403 Model 3! The -3 and N1 had the same raw speed and print head, but with the cover down the N1 was much quieter! I only ever saw a model 3 once. Until the last year or so I didn't know early -3s could attach to the IBM 7040 when I was researching 7090-7040 direct couple systems.

[steve-myers]

Going back to the topic: I'm pretty sure Dave Cole (zDBC I think) works this way, but I couldn't find anything in Google.