Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by postgresql.org (Postfix) with ESMTP id 986A5475461 for ; Sat, 6 Apr 2002 10:45:29 -0500 (EST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.11.4/8.11.4) with ESMTP id g36FjaF25654; Sat, 6 Apr 2002 10:45:36 -0500 (EST) To: d_nardini@btconnect.com Cc: pgsql-sql@postgresql.org Subject: Re: intersect performance (PG 7.1.3 vs 7.2) In-reply-to: <8F149658-4943-11D6-876E-0030654E696C@btconnect.com> References: <8F149658-4943-11D6-876E-0030654E696C@btconnect.com> Comments: In-reply-to d_nardini@btconnect.com message dated "Sat, 06 Apr 2002 09:49:16 +0000" Date: Sat, 06 Apr 2002 10:45:35 -0500 Message-ID: <25651.1018107935@sss.pgh.pa.us> From: Tom Lane X-Archive-Number: 200204/80 X-Sequence-Number: 7345 d_nardini@btconnect.com writes: > Following are EXPLAIN dumps from each platform for both intersect and > union (both return the correct data on each platform). Did you ever VACUUM ANALYZE these tables on the 7.1 box? The statistics estimates from that machine look suspiciously like default values. If you did, then maybe I can credit 7.2's improved statistics machinery with a win ;-). But I bet the better choice of plan on the 7.2 box is due simply to having any stats at all. Can't help you with the RPM dependency issues, sorry. regards, tom lane