On Wed, 2008-02-13 at 13:29 +0100, Dawid Kuroczko wrote:
> > * no direct table cache control;
>
> Could you elaborate more on this one?
Well, I was just summarizing what other people wrote :-)
But I guess they refer to table level control of how much cache memory
to use. I think there are DBMSs where you can tune the cache usage based
on tables/table groups, so you can make sure the important tables (where
you want instant response even if the table was not accessed in the last
hours) are always in cache. Currently that's not possible in postgres -
it is true that the cache memory will be used in an optimal way
regarding mean performance, but sometimes the mean performance is just
good enough even if not optimal, and you would like to eliminate worst
case scenarios for some queries. In that case more control (on the cache
strategy too) is better...
Cheers,
Csaba.