![]() |
![]() HP OpenVMS Systemsask the wizard |
![]() |
The Question is: We run a large OpenVMS cluster, and use RMS files that are accessed by many (>5000) processes concurrently. It is argued variously that: 1. A product like Oracle handles this more efficiently than does RMS 2. It would be better to have a few server processes directly accessing the files 3. The Lock Manager has too much overhead to allow us the throughput that we need, and that we should bypass it, letting the server processes manage concurrent access through some other approach. I realize that this is pretty vague, but I do not know where to find info to address these kind of issues, in particular, some reference that compares RMS performance to that of commercial DBs. The Answer is : Performance information depends entirely on the precise nature of your particular application, rather than the experiences of others. Start with the guidelines set forth in the OpenVMS Guide to Performance Management manual, and work from there. You should make use of the available performance monitoring tools to determine where the time is being spent and whether an alternate approach might even affect that -- the essence of tuning is locating the specific bottleneck and removing it. One common and obvious source of performance bottlenecks can be the application designl. Considering the cost of of reprogramming an application, another alternative might be hardware reconfiguration.
|