I'm currently setting up a Gunicorn application on a single core machine, and I have a limitation on memory so I would rather use 1 Worker and 8 threads than 2 workers with 4 threads.
However, I still asking myself about a possible performance issue with a 1 worker configuration as on the Gunicorn official documentation recommend using the following setting.
2 * (number of cores) + 1
Considering the fact that with Python I'll be natively blocked by the GIL If I want true parallelism using threads, I still don't see any interest using more processes than the number of core.