|
I have done simple performance test of clear IIS (without other ISAPI filters) and IISTracer. The tests was with several kind of applications
1. Short html file (1B) - this test shows maximum performance loss of IISTracer
2. Normal HTML file (16kB) - standard performance loss on normal static files
3. Simple ASP file - ASP file with one byte, without code
4. Simple application - Short code with one CreateObject and 16kB Response.write.
|
Processor time |
Requests type |
req/s |
Without IISTracer |
With IISTracer |
1B HTML |
420req/s |
10% |
16% |
16kB HTML |
250req/s |
15% |
24% |
1B ASP |
250req/s |
25% |
28% |
Simple ASP |
100req/s |
50% |
51% |
The results shows that the maximum theoretical performance loss is for very short HTML files - about 60%. Normal performance loss with IISTracer for static files with size is about 10-20kB is 30% - 40%.
ASP takes many time to its initialization - so performance loss with ASP requests has theoreticall maximum of 10%. Performance loss for normal ASP applications is unmeasurable (maximum is 1%-2%). The same situation will probably be with other applications, such as .cgi, .php or .idc.
Performance loss of IISTracer for normal applications - .asp, .cgi, .php with many commands - is unmeasurable (is very small). |