I like the discipline of publishing my results on a blog. It forces me to double check my results and look for discrepancies in data.
The numbers I got in my recent Taurus test seemed off. Using a modified example configuration (concurrency 50, time 300s,) Taurus was able to completely saturate 4 cores of VM it was runnig on. At the same time, it was not able to saturate the VM with the Nginx instance under test. Nginx’s VM was using on average 80% of its single core. Taurus generated 352 requests per second to main page, which resulted in 2812 requests per second if I count static dependencies that were being downloaded, since Taurus was ignoring cache settings. Those numbers agree with the Nginx access.log entries.
...