| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | Michael Glaesemann <grzm(at)seespotcode(dot)net> |
| Cc: | "Phoenix Kiula" <phoenix(dot)kiula(at)gmail(dot)com>, "PG-General Mailing List" <pgsql-general(at)postgresql(dot)org> |
| Subject: | Re: Timestamp indexes (why ">" or "between" does not use index?) |
| Date: | 2008-02-16 15:39:39 |
| Message-ID: | 19464.1203176379@sss.pgh.pa.us |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-general |
Michael Glaesemann <grzm(at)seespotcode(dot)net> writes:
> The planner will choose a seq scan if it thinks that it will be
> faster than using an index: if based on its statistics it thinks a
> large portion of rows will match the criteria, a seq scan may well be
> faster than an index scan.
> Have you analyzed recently?
If you've never analyzed at all, the default assumptions will
discourage the planner from using an indexscan for a one-sided
inequality condition (like "x > constant"). On the other hand,
it usually will use an indexscan for a range inequality (like
"x > constant1 and x < constant2", or a BETWEEN construct).
If you do have ANALYZE stats then it all depends on what fraction of
the column's range is selected by the inequality or range condition.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Tom Lane | 2008-02-16 15:47:24 | Re: SELECT CAST(123 AS char) -> 1 |
| Previous Message | Arturo Pérez | 2008-02-16 15:38:56 | nntp interface not working? |