--- On Tue, 11/2/10, Ted MacNEIL <[email protected]> wrote: From: Ted MacNEIL <[email protected]> Subject: Re: How long for SMF to switch To: [email protected] Date: Tuesday, November 2, 2010, 3:03 PM
>Our process to offload the SMF data from the MAN files waits until all MAN >files are full. >(Yes - I will be fixing that). I've never worked in an environment that did it that way, so I won't comment. >So it forces a switch SMF between each step. >Because the MAN files are small on very active systems >(Yes - I will be >fixing that), any time I can save in switching will reduce the amount of time >I am buffering SMF. I'd suggest that you do fix the two above. Then 1.5 minutes won't be an issue. It's a symptom -- NOT the problem. Ted: As you probably know every shop is different and there are really no hard and fast rules. I would research (get record count) off the smf tape for each dataset and try sampling say over 6 months time (make sure you get the window in year end or anyh other time there is a lot of activity) and size ithem an go from there. AT one shop I was in the programmers were terrible with COBOL. Example: when they wanted to go back to the first record they would close and re-open the FD (yikes) I happened to be out in the compouter room and noticed SMF address space was eating memory up so I had them start dumping. It never caught up as the program was creating the vsam open/close for evey record in the file. I finally had to do a force on the job. I read the programmer the riot act. I also called his boss squared to make sure that never happened again (it didn't). I suspect there were other similar programs flotaing around but did not have the volume og open/close that really hurt anything. If I had wanted to I could have gone back for a years worth of SMF to see but I had my hands in the burner for to many other items. I had the idea that the programmer did not have a clue on VSAM and got advice from someone else when he wrote the program. I asked that they send the programmers to class on COBOL but everyone was insulted that I asked.I am a fan of COBOL up to a point but frankly not enough education was not out there for typical joe programmer. IBM's cobol education was almost a joke but at that time Steve Comstock type was not out there or if he/she were they were silent. I had some input at that opoiny and the VP would have listened if had had a suggestion like Steve C. Ed ---------------------------------------------------------------------- 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
