Jan wrote:
> > The disk cache on most operating systems is optimized. Plus,
keeping
> > shared buffers low gives you more room to bump up the sort memory,
which
> > will make your big queries run faster.
>
> Plus, the situation will change dramatically with 7.5 where the disk
> cache will have less information than the PG shared buffers, which
will
> become sequential scan resistant and will know that a block was pulled
> in on behalf of vacuum and not because the regular database access
> pattern required it.
Hm. In my experience the different between data cached between shared
buffers and the O/S is not very much...both are fast. However, I almost
always see dramatic performance speedups for bumping up work mem. Are
you suggesting that it will be advantageous to bump up shared buffers?
Merlin
