Bruce Momjian escribió:
> > So, are you saying you like 4x now?
>
> Here is an arugment for 3x. First, using the documented 25% of RAM, 3x
> puts our effective_cache_size as 75% of RAM, giving us no room for
> kernel, backend memory, and work_mem usage. If anything it should be
> lower than 3x, not higher.
The other argument I see for the 3x value is that it is a compromise.
People with really large servers will want to increase it; people with
very small servers will want to reduce it.
> Finally, for those who like the idea of 4x, you can think of
> shared_buffers (1x) + effective_cache_size (3x) as totalling 4x.
This part of your argument doesn't work really, because AFAIR the
effective_cache_size value ought to consider that shared_buffers is part
of it (so e_c_s is shared_buffers + kernel cache). So if you're seeing
the 4x as e_c_s + s_b, you would be counting s_b twice.
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services
