Are the files going to use the Perform Script on Server script step or WebDirect? These can be important factors.
CWP, XML, or even long running server schedules might be a significant factor too, but not typically.
All of them. We've deployed on 24+ core machines
I presume that users will see less wait time if I go from my current config of 4 cores in FMServer 11, to 8 or 12 cores with FMServer 13. My hope is that more requests will get processed at the same relative time.
Yes, assuming that is where the FMS stats log shows that your current bottleneck is. If the
- elapsed time/call
- wait time/call
are high then they will see improvement
If however the delay is all in
- i/o time per call
then you may get bigger bang for the buck by improving the disk i/o
Perhaps you've seen something different, but my experience has been that processes (in the macro sense) that stay within the fmserver process use a single core. The additional cores have only come in to play when something can get "spun off", e.g. a script running under the fmsase process, backup process, etc. But if a user triggers a server-side sort, that will stay within the main fmserver core.
Granted, nearly all of my recent Windows FMS installs have been on esxi based VMs, but one of these had 24 cores allocated to it. I'm interested to know if the VMs somehow colored my assessment though.
I do see things happening differently. Both in VM and on physical boxes. It is true though that one core will be busier than the others, there is not a lot of load-balancing going on, more an "overflow" approach. But all cores will get busy with client requests if need be.