I have been searching for the optimal wrk2 parameters to maximize requests/second, so I ran some tests and got unexpected results. I've set up an Nginx web server putting out the default welcome page and ran wrk2 from another VM. The setup was pretty much as follows: 2 VMs - 16 cores, 8GB ram each, running Ubuntu 20.04.6 live server, both VMs are on the same physical machine running Proxmox.

The default command was: wrk -t 1 -c 1 -d 60 -L http://192.168.108.185:8080/ -R 1 -r

• each run I doubled R until reaching R > 131,072
• then R would be reset and I would double c, until c > 32,768
• then I would reset c and r and double t

In the results I got a clear pattern that setting R < t was giving an unexpectedly high R/s. If anyone has any idea why does it happen I'd love to know! If you have any questions about the test, I'd be happy to give you more information, below are some graphs I made after getting the result data:

• Target R/s is the -R parameter
• Threads is the -t parameter
• Connections is the -c parameter
• The colours on the heat map represent actual requests per second