| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | Együd Csaba (Freemail) <csegyud(at)freemail(dot)hu> |
| Cc: | 'Együd Csaba' <csegyud(at)vnet(dot)hu>, pgsql-general(at)postgresql(dot)org |
| Subject: | Re: Strange Index behavior |
| Date: | 2004-12-22 18:00:46 |
| Message-ID: | 5666.1103738446@sss.pgh.pa.us |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-general |
=?iso-8859-2?Q?Egy=FCd_Csaba_=28Freemail=29?= <csegyud(at)freemail(dot)hu> writes:
> The difference between the result times is 16x. I can't understand why the
> planner thinks it is the better way...
The planner thinks that because it thinks the second query will fetch
1500 times as many rows as the first. Now that I look more carefully
I see that both of those row estimates are off, the first by a lot.
Have you ANALYZEd this table lately? If you have, it might be a good
idea to increase the statistics target for the tstamp column.
You might also need to play around with random_page_cost, but it will be
counterproductive to touch that until the rowcount estimates are in the
vicinity of reality.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Scott Marlowe | 2004-12-22 18:08:10 | Re: postgresql.conf |
| Previous Message | A. Mous | 2004-12-22 17:58:07 | Re: 8.0 Beta3 worked, RC1 didn't! |