My experiences are this, all AIX using several Cloverleaf versions 3.3.0P through 6.0.1.
1. It seems to be more volume and processing related vs. actual numbers of processes per side, or number of threads per process.
2. There is a limitation (I know, I’ve seen it), where if you have too many threads/processes in a site, that the monitord doesn’t report the the GUI, showing things in incorrrect status or not allowing commands to be run.
Things to watch out for.
1. Cloverleaf can’t get messages out of state 5 fast enough for a particular process. This means the process is doing too much and needs to be split.
2. Cloverleaf threads that route to other Cloverleaf threads backup up in state 7 or 11. This means where your sending data is overloaded because it’s holding up the receipt of data. Same could be said if the source system is reporting high queue depths to Cloverleaf.
3. High volumes in recovery dB either because of #1, #2 or a destination system outage. This will slow everything down, to the point where the monitord cannot report on things, the GUI freezes, and recycling processes takes and extremely long time.
There are no hard or fast rules, it’s like tuning an older car, you have to get the feel for it, and identify the trouble signs via vigilant monitoring to prevent catastrophic failure.
Hope this helps,
Robert Milfajt
Northwestern Medicine
Chicago, IL