I had a customer of mine expand the number of virtual processors on their AF/Analysis server from 2 to 8, but when I look in the Analysis Service Statistics -> Manager Statistics, I see the AnalysisProcessorCount is 1.
Can I do anything to make the service utilize more processors?
The reason for increasing is that there are millions of analyses in the execution queue for natural triggered analyses, and this number is increasing. I have taken steps to get rid of the most heavy analyses (such as TimeNE(...., BOM('*'),'*', 0) is rather time scheduled every 5 mins, and only looks at the last 5 mins every time). We have a poor server environment with high latency between the data archive and the AF Server, so that's one of the causes for the massive queue.
The virtual server is running on VMWare and it seems to be possible to change RAM/CPU allocation while servers are running, because even though the cpu count has increased I see the uptime of the server is still over a month. That means the analysis service hasn't been restarted after the increase. This is good and bad - good because I fear I would loose the long execution queue if it restarted (I won't have to manually backfill) and bad because my suspicion is that the service isn't picking up on the improved hardware.
Any inputs on this?