Loadtest stats for parametrized use cases
I want to measure average response times of the separate calls that together make up a use case. But I want to run this use case for 10 different datasets.
What I ended up doing was define a very long testCase, containing steps:
step1dataset1
step2dataset1
...
step1dataset1
step2dataset2
...
which is hardly maintainable and very annoying: all step1 instances are the same exact for the data in the request. However, if I make a groovy loop over dataset-ignorant step 1 to 9, my statistics won't be fine grained enough: I need to see average response times over 20 runs of dataset 1, avg response times over 20 runs of dataset 2, etc. A groovy loop would result on avg response times of steps 1 to 9, not taking into account that dataset 10 takes more time to process than dataset 1
any suggestions? would it be inevitable to write my own statistics, using testResult.getTimeTaken() ?
What I ended up doing was define a very long testCase, containing steps:
step1dataset1
step2dataset1
...
step1dataset1
step2dataset2
...
which is hardly maintainable and very annoying: all step1 instances are the same exact for the data in the request. However, if I make a groovy loop over dataset-ignorant step 1 to 9, my statistics won't be fine grained enough: I need to see average response times over 20 runs of dataset 1, avg response times over 20 runs of dataset 2, etc. A groovy loop would result on avg response times of steps 1 to 9, not taking into account that dataset 10 takes more time to process than dataset 1
any suggestions? would it be inevitable to write my own statistics, using testResult.getTimeTaken() ?
