I didn't understand MVS that well until I took a VM performance class. It's been a long time and these are generalized comments. Others will probably comment as well....
The hardware is what makes the software happen, research the S/390 Dynamic Address Translation box, ie the DAT box. The hardware uses tables maintained by software to translate virtual to real addresses. The software can do creative things with these tables, like present each task (address space) with a table that contains some private storage and some shared storage. How is storage shared? The values for a virtual address point to the same real address in different address spaces. I think in MVS, storage is either private or "common" shared between all address spaces. In VM, any number of agreeable address spaces can have a shared storage area. Other address spaces could share a DIFFERENT storage area at the same virtual addresses. (VM Shared Segments) When a task starts, a limit of virtual storage is established. Certain control blocks are allocated, but not much user storage is allocated until it is requested via things like GETMAIN. Only allocated storage is backed by real and only paged out to aux as the system deems necessary. A task may allocate storage, but never use it. The system monitors the amount of storage a task uses frequently and established a "working set size". I just looked at my TSO session and it had a working set of about 5M even though I logged in with the max region size allowed. ---------------------------------------------------------------------- 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
