calculating effective band width used during a load test...
Using SilkPerformer we conducted performance test for a web application 3 times in a similar environment.
- load test 1 from 10 am to 11 am.
- load test 2 from 11.30 am to 12.30 pm.
- load test 3 from 1 pm to 2 pm.
For all three test we considered same parameters of load test project and testing environment.
What we found is, the response time calculated for various transactions in 3 load test differ a lot.
There is a lot of difference between response times of one load test and another load test.
I would like to know why this much of variation occured in response times.
In such cases how do we consider test results.
Is there a way to calculate effective band width used during a load test?