SMA performance optimisation question - time to start
I'm a bit disappointed by the average time to start for most of our runbooks. Our Production instance has 2 app servers load-balanced and we are not using the WAP at all. The average time to start for some runbooks is anywhere from 1:1 to 8:1 of the execution duration (time to start - time to finish).
I haven't looked into the volume of jobs over time yet, but runbook executions are <100 per day so I doubt it'd be overloaded. SMA servers are 8GB, 2 core @ 2.66GHz and never push their resources very hard.
I can't think of anything unique about the runbooks that start faster on average compared with those that start slower. Almost all create a remote session on the SCSM primary workflow server, and one of them is invoked by a Cireson PowerShell activity, while the others are invoked either by an SCSM workflow or a SMA RB activity. We have Log Verbose set on all runbooks because we are executing so few we aren't concerned about the associated DB growth.
However, I doubt any of that has an effect on the time to start; more likely the execution time.
Does anybody know how to speed this up?