> why the hell introduced IBM an exit like IEFUSI? There must be an > reason <<>> It's not because of missing planing, communication and > so on. It's just to ensure the health of operating system in case > of ......
"Back in the day..." the world was much less complex than today. Workloads were basically batch (homogeneous) and resources were much more constrained. Jobs could run for hours and fail when they ran out of storage. That wasted hours of precious processor time. It would have been better to fail the job at the start, than let it burn. That was the purpose of the REGION keyword and folks used to pay very close attention to it when I was in first grade. It made sense to provide a gross control over storage requests. That was IEALIMIT. Then came virtual storage and people got a whole lot less careful. IEFUSI came later. It was the same general concept, just spiffed up for 31-bit and then later, 64-bit operation. Basically, you take a look at what the programmer asked for in the JCL, compare that with your chicken entrail rules de-jure, add a dash of second guessing and adjust, allow, or reject the request. Even when it was brand new it was not exactly a brilliant concept. IEFUSI is to virtual storage management as pliers are to microsurgery. There are too many confusing/conflicting controls and IEFUSI doesn't really protect the system so much as just get in the way of getting things done. In theory, you could write a very rich IEFUSI that was cognizant of the requirements of every job in the system, but that would be absurdly complex in most situations and in any case, how would you deal with the inevitable exceptions and corner cases? What happens when a program that is legitimately allowed REGION=0M goes bonkers and eats storage? What happens when a "gotta run right now" job fails because it falls outside the IEFUSI programmer's conceptual infinity? And how do you change things on the fly when something has to happen right NOW? > If there is a known reason to change IEFUSI to set different limits > for different STC/JOB or whatever the sysprog will make the needed > changes via JES parms, SMF or IEFUSI So you're going to modify the IEFUSI exit for each needed change, reassemble it and activate it dynamically? Yeah right. And what happens when you're out on a smoke break when that SOX-compliance job just has to run right now and it keeps failing for lack of region? That's not exactly a responsive way to run a system. Depending on your business, it could be a lot more expensive than the extra DASD for AUX. Does anyone remember the SWIFT network? It was a money forwarding system for banks. If you didn't pass the buck within the required period, you owned the transaction. It could have been a dime or a billion dollars. Our management used to watch SWIFT responsiveness fanatically. You really did not want to be dinking with the system trying to get stuff to run while a billion dollar fund transfer was sitting in the hopper. > In IEFUSI we trust Not! ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
