Pierre-Frédéric Caillaud wrote:
> The planner is smarter with GROUP BY than with DISTINCT, so you can
> rewrite your query as the following, whihc will probaly use a
> HashAggregate, and be a lot faster :
>
> SELECT service_id FROM five_min_stats_200408 GROUP BY service_id;
Pierre-Frédéric, thanks! Vast improvement - this is definitely acceptable. It did indeed use a seq scan followed by a
hashaggregate.
Total runtime: 76295.775 ms
>> estat=> explain analyze select distinct(service_id) from
>> five_min_stats_200408; QUERY PLAN
>> ------------------------------------------------------------------------
>> ------------------------------------------------------------------------
>>
>> Time: 1284213.359 ms
--
Guy Rouillier