| From: | Markus Schaber <schabios(at)logi-track(dot)com> |
|---|---|
| To: | PostgreSQL Performance List <pgsql-performance(at)postgresql(dot)org>, PostGIS Users Discussion <postgis-users(at)postgis(dot)refractions(dot)net> |
| Subject: | Re: [postgis-users] Bad query optimizer misestimation because of |
| Date: | 2005-02-03 13:59:25 |
| Message-ID: | 42022E3D.80108@logi-track.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-performance |
Hi, all,
Markus Schaber schrieb:
> As a small workaround, I could imagine to add a small additional column
> in the table that contains the geometry's bbox, and which I use the &&
> operator against. This should avoid touching the TOAST for the skipped rows.
For your personal amusement: I just noticed that, after adding the
additional column containing the bbox and running VACUUM FULL, the table
now has a size of 4 pages, and that's enough for the query optimizer to
choose an index scan. At least that saves us from modifying and
redeploying a bunch of applications to use the && bbox query:-)
Markus
--
markus schaber | dipl. informatiker
logi-track ag | rennweg 14-16 | ch 8001 zürich
phone +41-43-888 62 52 | fax +41-43-888 62 53
mailto:schabios(at)logi-track(dot)com | www.logi-track.com
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Richard Huxton | 2005-02-03 14:52:04 | Re: Tunning postgresql on linux (fedora core 3) |
| Previous Message | Richard Huxton | 2005-02-03 13:56:50 | Re: Tunning postgresql on linux (fedora core 3) |