Re: timestamped archive data index searches
От | Bruno Wolff III |
---|---|
Тема | Re: timestamped archive data index searches |
Дата | |
Msg-id | 20020721114103.GA6647@wolff.to обсуждение исходный текст |
Ответ на | Re: timestamped archive data index searches ("Stephen Birch" <sgbirch@hotmail.com>) |
Ответы |
Re: timestamped archive data index searches
|
Список | pgsql-general |
On Sun, Jul 21, 2002 at 07:32:22 +0000, Stephen Birch <sgbirch@hotmail.com> wrote: > I am still puzzled by the systems use of sequence scans. Using Tom's > suggestion, I am now able to get a reasonable response time on the 1M > record database by searching on the tstamp field. > > But ... I tried asking the database what the earliest record is: > > SELECT MIN(tstamp) FROM det; > > This used a sequence scan even if I do a SET ENABLE_SEQSCAN to off. > > Shouldn't this also use an index? No because there isn't hardcoded special knowledge about the min and max aggregate functions. This gets discussed on the lists pretty often so you should be able to find more detailed discussions in the archives. If there is a usable index on column of interest you should rewrite your query to use order by and limit. For example: select tstamp from det order by tstamp limit 1;
В списке pgsql-general по дате отправления: