On Fri, 10 May 2002, large scale wrote:
> Hi,
>
> I have two tables, one has 25000 rows and the other
> has 6.5 million rows.
>
> (25000 rows)
> table1
> (id text,
> start int,
> stop int)
>
> with seperate index on three individual fiels.
>
> 6.5 million rows
> table2
> (id text,
> start int,
> stop int)
>
> with seperate index on three individual fields.
We'll start with the standard questions: Have you
vacuum analyzed? What version are you running? (if
it's less than 7.2, you may want to see about
upgrading) If you do a set enable_seqscan=false;
what does the explain show then? I'd be interested
in know if 1024601931 is even remotely a valid number
of rows from that join as well (which is about
.5% of an entire cartesian join if my math is right).
Perhaps some exists style thing would be faster since
that would at least presumably be able to stop when
it found a matching table2 row for a particular table1
id.