To: arnaulist@andromeiberica.com, pgsql-performance@postgresql.org
Subject: Re: Advise about how to delete entries
References: <43183AC9.1060704@andromeiberica.com>
Message-ID: <op.swhh87t5th1vuj@localhost>
Date: Fri, 02 Sep 2005 18:47:21 +0200
From: PFC <lists@boutiquenumerique.com>
Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?=
Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
In-Reply-To: <43183AC9.1060704@andromeiberica.com>
User-Agent: Opera M2/8.0 (Linux, build 1095)


>     "DELETE FROM statistics_sasme WHERE statistic_id = 9832;"

	As Michael said, why use a NUMERIC when a bigint is faster and better for  
your use case, as you only need an integer and not a fixed precision  
decimal ?

	Also if you use postgres < 8, the index will not be used if you search on  
a type different from the column type. So, if your key is a bigint, you  
should do WHERE  statistic_id = 9832::bigint.

	For mass deletes like this, you should use one of the following, which  
will be faster :

	DELETE FROM ... WHERE ID IN (list of values)
	Don't put the 30000 values in the same query, but rather do 300 queries  
with 100 values in each.

	COPY FROM a file with all the ID's to delete, into a temporary table, and  
do a joined delete to your main table (thus, only one query).

	EXPLAIN DELETE is your friend.