Hi Gatling Community,
I thought I would reiterate a question that I asked last week.
According to my knowledge it is an error in the measurement methodology to include the ramp-up and ramp-down periods when calculating the response time metrics. This error can be avoided by using filtering capabilities of a testing tool. I was searching for such capability in Gatling and I did not find one.
Have I missed something? Any pointers anyone?
If not then what is the approach of the Gatling community here? Is this somehow a ‘non-issue’? Is this a ‘non-issue’ just because the tool does not support that?
For sure, Gatling Enterprise manage ramp-up and ramp-down. But I’m not sure to what extend.
And I don’t remember (and have not the time right now) if the open source version does anything about that (I guess that is not a concern of the open source application).
I confirm this is featured in Gatling Enterprise, where you can define the time window of interest and ignore ramp ups and downs, in particular when computing assertions/acceptance criteria.
No problem and thank you for the answer. Anyway, you guys do a great job in supporting the community, as obviously the opensource community does not pay for the software. And I understand that the feature of filtering by time is only present in the enterprise edition.
Having said that, the fact that I am apparently the first person in the forum asking about this feature is rather puzzling. It looks like this built-in measurement error is widely accepted, which then raises a question about the quality of the performed tests. Well, possibly other challenges are more important.
In my case I will maybe see if the simulation.log can be filtered with a simple script and I will also suggest to the management to maybe have a look at the enterprise version :-).
Oh, and another similar feature is to remove the failed requests from the calculation of the metrics, it depends on the particular case, but it is usually a good thing to do.
Oh, and another similar feature is to remove the failed requests from the calculation of the metrics, it depends on the particular case, but it is usually a good thing to do.
This is also featured in Gatling Enterprise: there’s a status filter on the response time metric: All, only OK, only KO.