| From: | Manfred Koizar <mkoi-pg(at)aon(dot)at> |
|---|---|
| To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
| Cc: | jdunn(at)autorevenue(dot)com, "'Postgresql Performance'" <pgsql-performance(at)postgresql(dot)org> |
| Subject: | Re: index v. seqscan for certain values |
| Date: | 2004-04-15 23:04:24 |
| Message-ID: | 571u701m7fkc97oi1ul8qfuf8e94kvi8lg@email.aon.at |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-performance |
On Tue, 13 Apr 2004 13:55:49 -0400, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>Possibly the
>nonuniform clumping of CID has something to do with the poor results.
It shouldn't. The sampling algorithm is designed to give each tuple the
same chance of ending up in the sample, and tuples are selected
independently. (IOW each one of the {N \chooose n} possible samples has
the same probability.) There are known problems with nonuniform
distribution of dead vs. live and large vs. small tuples, but AFAICS the
order of values does not matter.
Servus
Manfred
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Manfred Koizar | 2004-04-15 23:32:53 | Re: query slows down with more accurate stats |
| Previous Message | Geoffrey | 2004-04-15 22:23:19 | Re: PostgreSQL and Linux 2.6 kernel. |