X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [64.117.224.130]) by svr1.postgresql.org (Postfix) with ESMTP id 86F4FD1C9F2 for ; Mon, 22 Sep 2003 20:43:43 +0000 (GMT) Received: from svr1.postgresql.org ([64.117.224.193]) by localhost (neptune.hub.org [64.117.224.130]) (amavisd-new, port 10024) with ESMTP id 98000-03 for ; Mon, 22 Sep 2003 17:43:13 -0300 (ADT) Received: from shire.ontko.com (shire.ontko.com [199.164.165.1]) by svr1.postgresql.org (Postfix) with ESMTP id 4DB85D1B4EB for ; Mon, 22 Sep 2003 17:43:11 -0300 (ADT) Received: from nick (bilbo.ontko.com [199.164.165.101]) by shire.ontko.com (8.12.3/8.12.3/Debian-6.6) with SMTP id h8MKhA5f027359 for ; Mon, 22 Sep 2003 15:43:13 -0500 Reply-To: From: "Nick Fankhauser" To: "Pgsql-Performance@Postgresql. Org" Subject: How to make n_distinct more accurate. Date: Mon, 22 Sep 2003 15:42:27 -0500 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200309/290 X-Sequence-Number: 3656 Hi- I have a table- called "event" with a field event_date_time that is indexed. There are 1,700,000 rows in the table and 92,000 distinct values of event_date_time with anywhere from 1 to 400 rows sharing the same value. (I did a count grouped by event_date_time & scanned it to get this info.) When I look at the pg_stats on this table, I always see 15,000 or lower in the n_distinct column for event_date_time. (I re-ran analyze several times & then checked pg_stats to see if the numbers varied significantly.) Since this is off by about a factor of 6, I think the planner is missing the chance to use this table as the "driver" in a complex query plan that I'm trying to optimize. So the question is- how can I get a better estimate of n_distinct from analyze? If I alter the stats target as high as it will go, I get closer, but it still shows the index to be about 1/2 as selective as it actually is: alpha=# alter table event alter column event_date_time set statistics 1000; ALTER TABLE alpha=# analyze event; ANALYZE alpha=# select n_distinct from pg_stats where tablename='event' and attname='event_date_time'; n_distinct ------------ 51741 (1 row) This number seems to be consistently around 51,000 if I re-run analyze a few times. I guess my question is two-part: (1)Is there any tweak to make this estimate work better? (2)Since I'm getting numbers that are consistent but way off, is there a bug here? (2-1/2) Or alternately, am I totally missing what n-distinct is supposed to denote? Thanks! -Nick --------------------------------------------------------------------- Nick Fankhauser nickf@doxpop.com Phone 1.765.965.7363 Fax 1.765.962.9788 doxpop - Court records at your fingertips - http://www.doxpop.com/