> One can't control their z/OS image, because the DASD for the RES is > controlled by the data center.
Right, one could not apply a patch to the nucleus. It is on a R/O volume. But you have pretty good control IMHO: - SYS1.PARMLIB/PROCLIB/etc. is your own. You can do anything you want there and IPL as many times as you want. - I suspect (but have never done it) you could fiddle with concatenations and so forth and have your own writable copy of some "OS-type" things. - You can have any current version of z/OS that you want, and stay with an old version as long as you want. - I get the feeling that IBM would work with you if you said "we can't tolerate PTF XX1234 just yet -- can you hold off for us?" But I'm not here to sell Dallas on behalf of IBM. Keep it in mind if $550/month is not too rich for your blood. But I was mostly describing it to give a possible baseline for discussions of "what would a mainframe co-op look like?" Charles -----Original Message----- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of ste...@copper.net Sent: Friday, July 3, 2020 12:15 PM To: IBM-MAIN@LISTSERV.UA.EDU Subject: Re: Mainframe co-op Years ago, in Silicon Valley, I worked on ACS/OBS WYLBUR. We had a P/390 that I had tuned the I/O for to really speed it up. ACS also sold time on their systems. Contractually, we were only allowed to charge access costs for the P/390. It was not to be a "production" machine. So developers could buy access to it, but not on a "per CPU time" charge and related. We did have a few takers for the P/390. The system Charles has mentioned has certain caveats and issues. One can't control their z/OS image, because the DASD for the RES is controlled by the data center. If one were to obtain a z/OS license, and were to get it to run under KVM, then one could have a "production" system, where all source is handled, compiles done, etc., while all system level testing is done on another image. There are costs with this that have to be overcome. Let's take a look into the future: IBM is going to put out a release of VM and/or z/OS that will not run on a z/?? CEC and that is the one you have (or SUSE/RHEL, etc. does the same with KVM etc.). You will now have to migrate to another machine. Can you get that machine on the used market at a good price? Meanwhile, you must have HLASM and probably want to have the toolkit (separately chargeable as I understand it). You will need all the compilers being used COBOL, PL/1, c/C++, etc.. Can you get them under a development license? Ok, let's say you can. You may need to have a small machine that is used for compiles so that you do not have to pay for the compilers on the bigger box. Given that you are going to have those who are doing development where they will need to have multiple CPUs, what you want is the slowest machine you can get (sub-model?) but with 4-6 General CPs for race condition testing. Now depending on the number of people/entities interested in this system, one may need multiple LPARs and possibly CECs to handle the workload. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN