Measure bottlenecks in IIS website performance?

Question

Knowing and using .NET memory and performance profiles (like RedGate ANTS Profiler) is very good in my development environment, I still haven't found a real way to do this on a production server.

Currently, I experience occasionally 100% w3wp.exe processor usage for a specific application pool.

Looking at the various performance counters, I still can’t determine the real reason, since I can’t determine in which part of my code the processor is under high load.

I tried CLR Profiler on a real server (with no meaningful result), and I could even imagine installing ANTS Profiler and let it run for several hours to get measurement logs.

Just to be sure that I am doing it right, my question is :

What is the best way to take some special measurements on the IIS 7 production web server to detect code bottlenecks?

Answer 1

Answering a question also on the Red Gate ANTS profilers forum, I received an excellent answer from the developers :

I basically pointed to the version of the Early Access Program (EAP) version 7 of ANTI Profiler. It has a feature that they call "continuous profiling" that basically performs direct tracing of the IIS workflow.

I will give this tool a try, as it looks like what I want to have.

Answer 2

Although you answered your question, I just wanted to mention Gibraltar .

I did not use it myself, but I heard well about it.

Answer 3

I will be tempted to accept something new in Relic APM - it will allow you to track both the real user interface (page load time), servers and the application itself

Take a look at this in more detail: http://newrelic.com/docs/dotnet/new-relic-for-net

Answer 4

Sam Shaffron (one of the creators of StackoverFlow) created a great command line tool some time ago, but unfortunately abandoned it.

My friend split the code to work in 2015:

https://github.com/jitbit/cpu-analyzer

(the page has a link to Sam's blog post explaining how to use it)