Date: Fri, 19 Aug 2005 20:11:54 +0200
To: "Tom Lane" <tgl@sss.pgh.pa.us>, "Bob Ippolito" <bob@redivi.com>
Subject: Re: sustained update load of 1-2k/sec 
Cc: "Mark Cotner" <mcotner@yahoo.com>,
	pgsql-performance@postgresql.org
References: <20050819082404.53751.qmail@web32915.mail.mud.yahoo.com>
	<67788A4A-4011-49AB-B329-683FD9532661@redivi.com>
	<17690.1124458553@sss.pgh.pa.us>
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
Message-ID: <op.svrot4mxth1vuj@localhost>
In-Reply-To: <17690.1124458553@sss.pgh.pa.us>
User-Agent: Opera M2/8.0 (Linux, build 1095)


> While I agree that hundreds of threads seems like overkill, I think the
> above advice might be going too far in the other direction.  The problem
> with single-threaded operation is that any delay affects the whole
> system --- eg, if you're blocked waiting for disk I/O, the CPU doesn't

	You use UDP which is a connectionless protocol... then why use threads ?

	I'd advise this :

	Use asynchronous network code (one thread) to do your network stuff. This  
will lower the CPU used by this code immensely.
	Every minute, dump a file contianing everything to insert into the table.
	Use another thread to COPY it into the DB, in a temporary table if you  
wish, and then INSERT INTO ... SELECT.
	This should be well adapted to your requirements.