Discrepancy between avg and tps for a single step test case
I am executing a simple strategy load test for a single step test case. I have 7 threads, 0 delay, and 0 random.
I get an average step time of 1s. This would suggest a tps of around 7, which is what I see if I set the calculation to work off the average.
However, using the elapsed time calculation I get a tps of 2.5, basically because the transaction count is much lower than it should be for a figure of 7 tps.
This suggests to me there are "hidden" delays in the system - either the measurement of the transaction time is too low, or there are significant internal delays within SoapUI in running each thread.
The only articles I've seen suggest the two tps calculations should be giving me roughly the same number.
Has anyone else seen this and got to the bottom of what is causing it?
I get an average step time of 1s. This would suggest a tps of around 7, which is what I see if I set the calculation to work off the average.
However, using the elapsed time calculation I get a tps of 2.5, basically because the transaction count is much lower than it should be for a figure of 7 tps.
This suggests to me there are "hidden" delays in the system - either the measurement of the transaction time is too low, or there are significant internal delays within SoapUI in running each thread.
The only articles I've seen suggest the two tps calculations should be giving me roughly the same number.
Has anyone else seen this and got to the bottom of what is causing it?