Many of us know the impact when globalmax hits 100%. Most places warn at 80% but depending on the globalmax definition, we can go from 80% to 100% rather quickly. Since this is a show stopper if there are problems, I was thinking about the use of the GLOBALUPDATE and GLOBALTEMPUPDATE parms to create some enhanced pro-active monitoring.
We could have a process that looks at the GLOBALUPDATE and GLOBALTEMPUPDATE counts every xx seconds. Get the previous count, the new count and look at the growth. If the counts (growth) exceeds some value, alert for awareness. Of course there would have to be some tending done initially so we can have some base line number to consider as an exception. This could help get us engaged faster and help prevent us hitting the 80% warning mark. It would be a early warning type mechanism. This could also help us identify old process that update a lot of globals that can be enhanced or decommissioned.
Has anyone attempted this proactive monitoring?
I was thinking maybe a new OPS/MVS product enhancement. Have the product alert like it does with messagerate and caienfrate (these are extremely beneficial). Also, maybe a parm that shows the last user, rexx or rule that did the update. This could help us identify who the culprit is. Maybe even show the last 5 users, rexx's or rules that did updates > xxxxx. This could really enhance proactive monitoring of the global environment so hitting the max is a very rare occurrence. We would have a safety net process for early awareness.
Just wanted to throw this out for thoughts and brainstorming.
Currently there is no further input from Development. They agree with Dave Gorisek's input on your posted idea. For the time being we would suggest that until we review the idea further. Thank you Jimmy.
idea and response to idea posted here.
Performance - New Global Variable Usage Warning Parms