Обсуждение: left outer join taking too long?

Have a general SQL question:

I have a script that does a left outer join
and I think it's taking longer than it should; but I
can not verify that (because I don't have anything
to compare it with - yes, I know, please bear
with me).

[snip script]
explain

select
        a.contract,
        a.mbr_num,
        a.mbrfname,
        a.mbrlname,
        a.mbradr1,
        a.mbradr2,
        a.mbrcity,
        a.mbrst,
        a.mbrzip,
        a.bu,
        a.class,
        a.product,
        a."group",
        a.phone,
        a.mbr_sex,
        a.county,
        b.pharm_copay,
        'P'::char(1) as primary_covg
from mbr a left join t_mbr_ben_spans b on
 (a.contract, a.mbr_num) = (b.contract, b.mbr_num)
;

[shaunn@hmp ]$ psql -U shaunn -d bcn -f ./bruce.sql
psql:./bruce.sql:27: NOTICE:  QUERY PLAN:

Merge Join  (cost=1081799.72..1088792.93 rows=1518781 width=237)
  ->  Index Scan using t_mbr_ben_spans_i on t_mbr_ben_spans b  (cost=0.00..3053.66 rows=51333 width=30)
  ->  Sort  (cost=1081799.72..1081799.72 rows=1518781 width=207)
        ->  Seq Scan on mbr a  (cost=0.00..73602.81 rows=1518781 width=207)

EXPLAIN

[/snip scritp]

As I look at this, I'm led to believe that 'cost' will make this thing
take a few days and I don't know how to make it more efficient.

What am I doing wrong?

Thanks!

-X

"Johnson, Shaunn" <SJohnson6@bcbsm.com> writes:
> Merge Join  (cost=1081799.72..1088792.93 rows=1518781 width=237)
>   ->  Index Scan using t_mbr_ben_spans_i on t_mbr_ben_spans b
> (cost=0.00..3053.66 rows=51333 width=30)
>   ->  Sort  (cost=1081799.72..1081799.72 rows=1518781 width=207)
>         ->  Seq Scan on mbr a  (cost=0.00..73602.81 rows=1518781 width=207)

Increasing sort_mem would make that cost estimate drop.  What effects
it'd have on the actual runtime is harder to say; but I never trust
EXPLAIN's estimates very much ;-)

            regards, tom lane