en English

Re: [PERFORM] Bad n_distinct estimation; hacks suggested?

в обсуждении

Список

Период

Сортировка

От:	Tom Lane
Тема:	Re: [PERFORM] Bad n_distinct estimation; hacks suggested?
Дата:	24 апреля 2005 г. , 04:49:20
Msg-id:	25382.1114318139@sss.pgh.pa.us (см: обсуждение, исходный текст)
Ответ на:	Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Josh Berkus)
Ответы:	Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Andrew Dunstan) Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Marko Ristola) Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Simon Riggs)
Список:	pgsql-hackers

Показать дерево обсуждения

Скрыть дерево обсуждения

Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Josh Berkus, 23 апреля 2005 г., 23:34:12 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Josh Berkus, 23 апреля 2005 г., 23:48:12 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? ("Andrew Dunstan", 24 апреля 2005 г., 00:33:20 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Tom Lane, 24 апреля 2005 г., 04:49:20 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Andrew Dunstan, 24 апреля 2005 г., 17:58:46 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Josh Berkus, 24 апреля 2005 г., 18:31:49 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Josh Berkus, 24 апреля 2005 г., 19:09:10 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Tom Lane, 25 апреля 2005 г., 04:59:49 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Marko Ristola, 24 апреля 2005 г., 20:50:42 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Simon Riggs, 26 апреля 2005 г., 20:35:54 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Josh Berkus, 26 апреля 2005 г., 20:40:44 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Andrew Dunstan, 26 апреля 2005 г., 21:42:02 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Simon Riggs, 25 апреля 2005 г., 09:01:30 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Tom Lane, 25 апреля 2005 г., 15:23:42 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Simon Riggs, 25 апреля 2005 г., 18:53:12 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Josh Berkus, 25 апреля 2005 г., 19:08:21 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Josh Berkus, 25 апреля 2005 г., 19:13:25 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Andrew Dunstan, 25 апреля 2005 г., 20:43:30 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Mischa Sandberg, 27 апреля 2005 г., 05:38:16 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Andrew Dunstan, 27 апреля 2005 г., 13:44:13 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Josh Berkus, 27 апреля 2005 г., 15:26:16 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Mischa Sandberg, 28 апреля 2005 г., 15:21:48 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Markus Schaber, 03 мая 2005 г., 16:36:16 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Mischa Sandberg, 03 мая 2005 г., 21:33:18 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Josh Berkus, 03 мая 2005 г., 21:42:36 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Josh Berkus, 04 мая 2005 г., 00:51:51 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Mischa Sandberg, 04 мая 2005 г., 02:52:24 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (John A Meinel, 04 мая 2005 г., 13:30:07 )
Re: [PERFORM] Distinct-Sampling (Gibbons paper) for Postgres (Josh Berkus, 29 апреля 2005 г., 05:23:08 )
Re: Distinct-Sampling (Gibbons paper) for Postgres (, 29 апреля 2005 г., 13:45:35 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Tom Lane, 25 апреля 2005 г., 21:11:24 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Simon Riggs, 26 апреля 2005 г., 21:06:20 )
Re: [PERFORM] Bad n_distinct estimation; hacks suggested? (Gurmeet Manku, 27 апреля 2005 г., 13:43:18 )
Citation for "Bad n_distinct estimation; hacks suggested?" (Gurmeet Manku, 02 мая 2005 г., 17:32:59 )

Josh Berkus <> writes:
> Overall, our formula is inherently conservative of n_distinct.   That is, I
> believe that it is actually computing the *smallest* number of distinct
> values which would reasonably produce the given sample, rather than the
> *median* one.  This is contrary to the notes in analyze.c, which seem to
> think that we're *overestimating* n_distinct.

Well, the notes are there because the early tests I ran on that formula
did show it overestimating n_distinct more often than not.  Greg is
correct that this is inherently a hard problem :-(

I have nothing against adopting a different formula, if you can find
something with a comparable amount of math behind it ... but I fear
it'd only shift the failure cases around.

            regards, tom lane

В списке pgsql-hackers по дате сообщения:

Предыдущее:

От: Oleg Bartunov
Дата: 24 апреля 2005 г., 04:41:41
Сообщение: Re: Wierd performance issue with 8.1cvs

Следующее:

От: Thomas Hallgren
Дата: 24 апреля 2005 г., 06:05:24
Сообщение: Re: Wierd performance issue with 8.1cvs