Inflated numbers when profiling a .asmx web service
dabstar
Posts: 2
Forum,
I am profiling an ASP.NET application that consumes a .net web service.
Both the ASP.NET app and the Web Service is on my local computer and I have the source code available.
First of all, while profiling I experience page load times of my application that are 3-5 times as long as when i run the app without profiling. Is that normal?
Secondly, when looking at the profile results it seems that the webservice call on the asp.net side takes 1.5 times longer than the actual time being spent in the web service method. Is this an overhead of the profiler (switching between threads and contexts) or does the .NET web service implementation really provide that much of an overhead?
Thanks in advance for your help,
dabstar
I am profiling an ASP.NET application that consumes a .net web service.
Both the ASP.NET app and the Web Service is on my local computer and I have the source code available.
First of all, while profiling I experience page load times of my application that are 3-5 times as long as when i run the app without profiling. Is that normal?
Secondly, when looking at the profile results it seems that the webservice call on the asp.net side takes 1.5 times longer than the actual time being spent in the web service method. Is this an overhead of the profiler (switching between threads and contexts) or does the .NET web service implementation really provide that much of an overhead?
Thanks in advance for your help,
dabstar
Comments
I'm not sure. ANTS Profiler does cause your application to run more slowly, but that is compensated for by an internal algorithm.
I think the issue here could be that because ANTS Profiler restarts IIS, the first invocation of the webservice is going to be much slower than usual, anyway. Do you think this explains things?