We run Postgres/Drizzle in production and tuning params (shared_buffers, work_mem, etc.) has always been trial-and-error for us. Curious what workload characteristics it uses to auto-tune — read/write ratio, connection count, something else? Also curious how it compares to just running pgtune once and calling it done.
I'm very interested in something like this for MySQL 8.x, will be checking your future work on it.
We run Postgres/Drizzle in production and tuning params (shared_buffers, work_mem, etc.) has always been trial-and-error for us. Curious what workload characteristics it uses to auto-tune — read/write ratio, connection count, something else? Also curious how it compares to just running pgtune once and calling it done.