gatling performance

If we run a gatling test and gatling is running on the same host we have many spikes in the TPS trendline. Gatling sometimes use 40% CPU time, at the same moment TPS line is going down. If we move the gatling tool to a seperate server the TPS-rate is more a flat line.

How can we know what gatling is doing? Is this normal behavior?


Gert Jan

Impossible to say without actually monitoring the Gatling JVM.
Anyway, have both Gatling and the system under stress on the same host is not a good idea as they will compete for resources, like CPU.