There are 4 configurations to enable the parallel and do the optimization, but the documentation of PostgreSQL doesn't says anything about values or calculation. My questions are:
1- How to calculate the values of
max_parallel_workers,max_parallel_workers_per_gatherandmax_worker_processes?2- The
work_memcan be calculate on base of connections and memory(RAM), but thework_memneeds to change something if I enable the parallel?
My supposition is: if the machine has 8 cores the max_parallel_workers is 8 and the values of worker process and per gather are 32(8*4), the number 4 I took from the original configuration that is 4 gathers per 1 parallel work.
After some searching I found some answers, this can help who wants to enable and to have a base configuration, case you have 4 cores(CPU):
your max worker processes will be the amount of cores and the max parallel needs to have the same amount:
max_worker_processes = 4
max_parallel_workers = 4
the gather is more complex because this value can be manipulated base on your needs and resource it is necessary to test to get a best value, but to startup values you can use cores/2.
max_parallel_workers_per_gather = 2
This is not a final answer, there some missing points... I am still searching and updating this answer or waiting for a better one.