I guess one lesson not explicitly mentioned is that doing benchmarking on something like macOS can have lot of pitfalls compared to systems more intended for this kind of workload (Linux...).
But the mention of wrk immediately stood out as a red flag; there is notable problem with wrk called "coordinated omission":
The default limits of MacOS is not really web server benchmark friendly.
ulimit -a
Maximum size of core files created (kB, -c) 0
Maximum size of a process’s data segment (kB, -d) unlimited
Maximum size of files created by the shell (kB, -f) unlimited
Maximum size that may be locked into memory (kB, -l) unlimited
Maximum resident set size (kB, -m) unlimited
Maximum number of open file descriptors (-n) 256
Maximum stack size (kB, -s) 8176
Maximum amount of CPU time in seconds (seconds, -t) unlimited
Maximum number of processes available to current user (-u) 5333
Maximum amount of virtual memory available to each process (kB, -v) unlimited
Debian's limits are better:
~> ulimit -a
Maximum size of core files created (kB, -c) 0
Maximum size of a process’s data segment (kB, -d) unlimited
Control of maximum nice priority (-e) 0
Maximum size of files created by the shell (kB, -f) unlimited
Maximum number of pending signals (-i) 30744
Maximum size that may be locked into memory (kB, -l) 993336
Maximum resident set size (kB, -m) unlimited
Maximum number of open file descriptors (-n) 64000
Maximum bytes in POSIX message queues (kB, -q) 800
Maximum realtime scheduling priority (-r) 0
Maximum stack size (kB, -s) 8192
Maximum amount of CPU time in seconds (seconds, -t) unlimited
Maximum number of processes available to current user (-u) 30744
Maximum amount of virtual memory available to each process (kB, -v) unlimited
Maximum contiguous realtime CPU time (-y) unlimited
For testing I always create an instance somewhere (mostly AWS) and configure the OS to have much higher performance limits to enable better performance outcomes.
I guess one lesson not explicitly mentioned is that doing benchmarking on something like macOS can have lot of pitfalls compared to systems more intended for this kind of workload (Linux...).
But the mention of wrk immediately stood out as a red flag; there is notable problem with wrk called "coordinated omission":
https://github.com/wg/wrk/issues/485
https://github.com/giltene/wrk2
https://news.ycombinator.com/item?id=34148502
Benchmarking is difficult. And there is lot of subtlety on how you set up the client(s) and servers that can impact the results a lot.