We have an IIS 7.5 WAS hosted (net.tcp) internal WCF Service. It is running on a Win2008 X64 R2 server with 8 cores (16 with hyperthreading) and 16gb RAM. The App Pool is set to 16 max worker processes. We are trying to tune the client / service
so that we can run as many worker processes as possible and keep the server as close to 100% CPU utilization as possible. The client is coded for for 17 concurrent threads / WCF service requests (using .NET 4.0 TPL library). However, I believe
there is some bottleneck going on as the server CPU's never get above 40% utilization and the number of worker processes stays fairly low. What is the best way to determine where the bottleneck is? How do you determine if the bottle neck is client
side or server side? What perf counters should I be looking at client side? What perf counters should I be looking at server side? Is there a more efficient binding than Net.Tcp?
View Complete Post