Because NLog Async + Serilog Async discards on overflow by default (can be changed), then you are mostly testing the overhead from overflowing the async-in-memory-queue (and how fast it can discard logevents).
You might also want to test how the logging frameworks handles multiple threads logging concurrently (ex. 4 threads), which is normal in a standard web-application.
Notice NLog has a "batching" feature for the NLog-Console-Target that can be enabled with writeBuffer="true", that works very well with async="true". See also: https://github.com/NLog/NLog/wiki/Console-target
Interesting and good to know!
I will look further when I have spare time.
Meanwhile, if you are interested to do the fix or add more changes, don't hesitate to open a PR :)
Because NLog Async + Serilog Async discards on overflow by default (can be changed), then you are mostly testing the overhead from overflowing the async-in-memory-queue (and how fast it can discard logevents).
You might also want to test how the logging frameworks handles multiple threads logging concurrently (ex. 4 threads), which is normal in a standard web-application.
Notice NLog has a "batching" feature for the NLog-Console-Target that can be enabled with
writeBuffer="true"
, that works very well withasync="true"
. See also: https://github.com/NLog/NLog/wiki/Console-target