Re: Confusing Query Performance

On Wed, 1 Oct 2008, Gauri Kanekar wrote:
> "new_table1" is 18% of the the whole "table1".

>    ->  Nested Loop  (cost=186.26..647160.32 rows=42543 width=16) (actual time=655.832..6622.011 rows=5120582 loops=1)

>    ->  Nested Loop  (cost=0.00..414246.81 rows=25155 width=16) (actual time=19.578..4922.680 rows=5120582 loops=1)

The new table may be that much smaller than the old table, but you're
selecting exactly the same amount of data from it. The data is fetched by
indexes, which means random access, so the overall size of the data that
you don't fetch doesn't make any difference.

Matthew

--
Existence is a convenient concept to designate all of the files that an
executable program can potentially process. -- Fortran77 standard