How many server workers per microservice - Gunicorn with Uvicorn in FastAPI

Question

I'm not sure how many workers should I declare per every FastAPI based microservice. Some information I found here: gunicorn - how many unicorn workers do I have to have in production

As I understand number of workers should be:

number_of_workers = number_of_cores x num_of_threads_per_core + 1

So if I have 4 services based on FastAPI and my processor have 4 cores and 8 threads I should divide workers per service? Like for example:

number_of_workers = 4 x 8 + 1
number_of_workers = 33

fastapi_service_1 - 9 workers
fastapi_service_2 - 8 workers
fastapi_service_3 - 8 workers
fastapi_service_4 - 8 workers

Answer 1

fastapi_service - in this case, 4 workers