Message-ID: <4C23645E.5070101@2ndquadrant.com>
Date: Thu, 24 Jun 2010 09:57:50 -0400
From: Greg Smith <greg@2ndquadrant.com>
User-Agent: Thunderbird 2.0.0.24 (X11/20100317)
MIME-Version: 1.0
To: Janning <ml@planwerk6.de>
CC: pgsql-performance@postgresql.org
Subject: Re: Write performance
References: <201006241443.33296.ml@planwerk6.de>
In-Reply-To: <201006241443.33296.ml@planwerk6.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

As others have already pointed out, your disk performance here is 
completely typical of a single pair of drives doing random read/write 
activity.  So the question you should be asking is how to reduce the 
amount of reading and writing needed to run your application.  The 
suggestions at 
http://wiki.postgresql.org/wiki/Tuning_Your_PostgreSQL_Server address 
that.  Increases to shared_buffers and checkpoint_segments in particular 
can dramatically reduce the amount of I/O needed to run an application.  
On the last server I turned, random reads went from a constant stream of 
1MB/s (with default value of shared_buffers at 32MB) to an average of 
0.1MB/s just by adjusting those two parameters upwards via those guidelines.

If you haven't already made large increases to those values, I'd suggest 
starting there before presuming you must get a different disk setup.

-- 
Greg Smith  2ndQuadrant US  Baltimore, MD
PostgreSQL Training, Services and Support
greg@2ndQuadrant.com   www.2ndQuadrant.us