Re: performance for MIN,MAX aggregates

On Tue, Mar 11, 2003 at 15:42:41 +0700,
Ruslan A Dautkhanov <rusland(at)scn(dot)ru> wrote:
> Hello all,
>
> Just a little simple example:
>
> isbs=# \d radauth
> Table "public.radauth"
> Column | Type | Modifiers
> ----------+-----------------------------+-----------
> dttm | timestamp(0) with time zone |
> username | text |
> realm | text |
> logline | text |
> Indexes: radauth_dttm_username btree (dttm, username)
>
> isbs=# explain select min(dttm) from radauth;
> QUERY PLAN
> --------------------------------------------------------------------
> Aggregate (cost=2591.75..2591.75 rows=1 width=8)
> -> Seq Scan on radauth (cost=0.00..2363.00 rows=91500 width=8)
> (2 rows)

This is discussed a lot in the archives. The short answer is use the
following query instead:
select dttm from radauth order by dttm limit 1;