We've had trouble with our DNN web/db configuration, which we initially solved by adding more RAM (16gig) and a 2nd CPU to the SQL server.
While we are stable now, we wanted to work on improving performance further, so we ran SQL Trace against the DNN db with the tuning template (Trace file stored off Server). This brought the web system to its knees. The trace log had roughly 144 transactions per second over the period of 10 minutes. Once we stopped the trace logging, the website came back to life - but here's what has us scratching our heads:
During the trace - SQL Server, CPU never above 60% and RAM never above 9Gig (both are allocated to as much as necessary). I don't know about Disk I/O. Web Server, CPU never above 80%, RAM never above 80% Disk I/O 90%, Network in the ball park of 100% - but we supposedly have a gigabit connection between the virtual servers (is seems illogical that we could be pegging that).
After turning off the trace - Web Server CPU and RAM pegged @ 100% - which to me suggested a back log that was trying to catch up, which has us leaning towards some type of queuing issue.
Keeping this open ended - what should we be monitoring to pinpoint where the bottle neck is?
Thanks in advance,
Chris