First, simply from your reply below, you are already probably doing everything exactly right. In a GRS STAR environment, I would expect a SYSTEMS ENQ to outperform a DASD RESERVE. Given that you are already doing everything right, I would guess the below won't help, but just in case....two more thoughts:
1) I would look at the RMF data for the ISGLOCK structure. Is the ISGLOCK structure wildly too small? If you see a large percentage of false contention, then the ISGLOCK structure is too small. What's too a percentage for false contention? Maybe 1% false contention, maybe 2%... False Contention is extremely expensive in GRS STAR mode. It is the only function in GRS STAR which is slower than the alternatives.... 2) What's the response time for the coupling links? Is it what it should be? Maybe there's something wrong there... If none of these help, then I'm all out of ideas, and you're probably back to talking to the vendor for assistance. Brian On Fri, 12 May 2006 04:45:07 +1000, George Kozakos <[EMAIL PROTECTED]> wrote: >Thanks to all that have responded. Just to clarify the environment it is >GRS STAR with ISGLOCK in a CF with a dedicated CPU. We changed from >hardware reserve + SYSTEM enqueue to global SYSTEMS ENQ. We now get >performance problems with enqueues issued via APPC for exclusive access to >SYSVSAM/LOGONIDS > >Regards, >George Kozakos ---------------------------------------------------------------------- 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
