> On Oct 15, 2017, at 9:22 AM, scott Ford <[email protected]> wrote: > > Ed, > > I agree with you. Machines are faster and faster. But customers are demanding > changes to code and fixes faster. This to me is a problem, especially if > 'caution is thrown to the wind'. Dealing with the security subsystem, I.e.; > RACF, ACF2 and TOP-SECRET , many customers are ignorant of these products , > especially in how and why they work. > > I write in COBOL and my first love Assembler, so we have exits that support > our product. > I have seen like everyone else a lot of craziness. The approach of using the > PC and Java for everything to me is a cop out . This is the apparent trend I > have been seeing. To this t-Rex this is not the only path to product > development. > > > Scott
Scott: Not only are machines faster the OS is in some aspects faster and faster. Just to give one example. We were one of the bloody early adopters of MVS. 1. Our two biggest issues were applications and how to speed them up. 2. Trying to account for the application code (who to charge) 2.5 Trying to keep MVS up long enough to get some production done (we had typically 6-12 IPL’s a week) Early on our SE (who functioned more of a high-level sysprog than an IBMer) discovered one of the big bottle necks was I/O. He wrote an orange book on it, and it was an eye-opener. One of the things he showed in the orange book was that when you add buffers to pretty much any QSAM file. The application would run faster. Thus was born SAMe (or was it samE). Approximately one year after the book was published IBM announce SAM-E. Since he helped write the code we were first (or close to) adapters. It worked miracles our run times decreased dramatically. Its been 30++ years so I don’t have the numbers in my head. But, for a $60 a month product it was a godsend. We had a three different groups that used our data center. The thought originally for the three of them to split the cost. After 5 or so years that wasn’t proving feasible, but we were damned to figure out who was getting the most. Out of the clear blue a box, about 1/2 the size of a 3830 controller showed up in the computer room. We had no idea what it was for; it turned out to be a solid-state drive for paging. We didn’t know who ordered it. The VP came down and announced he had (without telling or asking anyone). We looked at each other and said, but it wasn’t big enough for our paging volumes. But we had to show that we were a team and we decided to put PLPA on it. SO the after it was hooked up to a couple of channels and put it online, meanwhile several thousand people were doing data entry work, and we were expecting the system to crash. It didn’t so I went ahead and defined a VTOC and a PLPA on it. The next Sunday it was going live, and we crossed our fingers. Sunday went by and then Monday, then Tuesday we had a power glitch, and of course, the system crashed. When we tried to IPL, we couldn’t because the idiots who designed the box never dreamed of a power outage and we had to define a PLPA someplace else and IPL again to use the primary PLPA. The damn box lost the VTOC and the PLPA. So I scuttled around and did that and set it up in parmlib who when we IPL’d it would automatically come up and use it. Then we had another power glitch. This time it remembered the VTOC but not PLPA. So here we go again. It took five outages before we crated the damn product up and shipped it out. A year later another one of those mysterious boxes showed up. The VP came down with another person who turned out to be the salesman. The salesman was heard to say that he would guarantee this “baby” would find out why we couldn’t account for about 25-30 percent of the machine. The next day the tech person from the manufacturer showed up and started to open the box. There was nothing special about the box other than it had a few gauges on the front and what looked like a mini tape drive on the front. As he was unpacking, I asked exactly what the machine did, and he started going off on a sales routine. I cut him off after a few minutes and said how exactly it was to work. He pushed me off on a few more sales type answers. He started to bring long wires out, and I told him to stay away from the console, and he continued stringing more and more wire out. He motioned to go closer to the console, and I said you are not going to touch that machine. We have several thousand people working and cannot take any outages. He assured me that he wasn’t going to do anything that would bring down the system. I finally told him to STOP, and I reached for the phone for the VP. He said OK, when can I hook this up. I said we would have to talk with the shift supervisor as I do not know the schedule for the building and our connections in Europe and the east coast and west coast. We went over to him and asked when the next open slot was where we could have the machine for two hours. He looked over his notes and said how on Sunday between 6 and 8 pm. I said OK. AO Sunday night I go down to the DC, and he was waiting for me, and I asked if he was ready, he said I was ready three days ago. Anyway, he hooked his apparatus up to the console, and he was fiddling around. After 30 minutes he said he was done, I walked over to the supervisor (managers don’t work on Sunday) and I said OK the machine is all yours. So he had the operators fire up some initiators, and the normal jobs ran, etc.. He told me that he would be back in a week and I said what if this causes issues, he said but can’t its passive. So we go though an entire month without issue. The last day he comes up to me and says can I dismount the tape, and I said go for it. He mounts another tape and leaves. SO fast forward a week and the salesman is supposed to give a presentation as to what they found. He gets up before a MIC and starts in about the business etc.… I raised my hand and asked if he can get to the point. He starts with a cough and rustling his notes, and he said well this was the first for him and he said there must have been something wrong with the machine as it didn’t show anything but 98 percent cpu utilization. I raised my hand and asked how the machine monitored everything. He got his expert up there and told us that the machine monitored key 8 and other keys as well. I looked at him, and said are you familiar with MVS and he said well no it's not widely used. I told him that every batch job used KEY 8 and Vtam and the others used their various keys, and your monitor is useless on an MVS system. You should have seen the expression on his face it was priceless. Everybody got up to leave, and I asked how much do you charge for this box? He said 100K quite a few of us laughed and left the room. Fast forward another year and the political heat up in corporate was getting intense. So to try and placate them our VP brought in a software package from Duquesne Systems called QCM. It claimed the same thing (find out where part of our system was going). We installed it, and there were lots of bells and whistles we found that was really useful. We were comfortable with it, but we still couldn’t find the lost CPU processing. After five months we could not find it. I was going through the reports, and something didn’t look right. I started to drill down and finally almost accidentally found it. One of our users was a big time user of execution batch monitoring and clerks would sit at their terminals and do what ifs and these what if's were programs that were nothing more than macros that assembler H took and created a load module that was later used in batch processing. The assemblies were being dropped by the reporting program and were not accounted for. Well, when that hit corporate the lid came off the roof. Apparently, that was enough for the one part of the company to be put on the market. Two+ years and countless hours and many outages later, that was the net result. The company ended up in Tampa, and the “rogue” CPU hog was sold off to another company. Ed ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
