Add up the ‘maxXXXconnections’ for each instance. add 20 (default parallelism) for the etl. iirc 1 process = 1 connection and 1 session = 1.1*processes or something like that.
There is a 1:2 relationship between the maxwit and maxjobexecution connections, need to look up some of the others but you can’t just reduce the maxXXXconnection settings w/o affecting other things.
How many appserver instances do you have ?
We have 4 app servers, all set with the same max connection settings and WITs...
In case this helps, this is what we have for max connection related settings:
So assuming they are all ‘ALL’, and the default settings that’s 300x4 + 20 + (spawner should be 100 max)x4
Now – rarely have i seen a customer max out the client or general connection pools. Certainly the job pool. What is maxwit and maxjobexecutionconnections set to on these ?
ok, if you take into account, the settings I've edited in my comment above, and the fact we don't spawn externally for now (we don't use the external spawner, in case that makes a difference), and that we use the NSH Proxy and have quite a bit of NSH Script Jobs running daily at all times, does that change?
From your calculation above, it would need to be set to 1,620 max connections on the DB server to handle the load?
That seems about right. you could get the dba to bump to the max and then setup monitoring on the pool usage and adjust down if you need to.
Alright, we'll start with that then. Thanks for the hints !