Switching from jmeter to hyperfoil
Closed this issue · 4 comments
In the past few months, we have seen countless broken benchmarks. Most of them were blindly using jmeter without understanding coordinated omission problems. Basically: do not trust jmeter.
It would be great to switch to hyperfoil (which we can trust) instead of jmeter to avoid such a problem and have a place showing how to do it.
I'm sure @franz1981 would be happy to help on this one (as he has to explain the problem every time).
Thanks @cescoffier very true!
@johnaohara and @edeandrea are already working on a version of this that's using hyperfoil (yay!)
- I can happily review both results and benchmarking scripts for sure
@edeandrea would you be able to make the change after Devoxx UK?
@geoand it's not as simple as just "make the switch" :) but yes it's being worked on (when there's time) :)
I know, but it's important to get Quarkus users the message that JMeter is not to be relied on for load testing - we need to be dogfooding our own message