Re: INDEX and JOINs

Il Friday 26 October 2007 15:18:04 Tom Lane ha scritto:
> Reg Me Please <regmeplease@gmail.com> writes:
> >>> (cost=3.95..382140.91 rows=274709 width=91) (actual
> >>> time=1.929..57713.305 rows=92 loops=1)
> >>> Hash Cond: (t_dati.camp_id = t_campi.camp_id)
>
> I think the reason it doesn't want to use an indexed join is the large
> estimate of the number of join result rows.  You need to try to get that
> number down to something nearer the reality.  Increasing the statistics
> target for the larger table might help.
>
>             regards, tom lane

How can I "Increasing the statistics target for the larger table"?
I'ìm sorry for asking, but I'm not that deep into RDBMS.

Thanks.