Message-ID: <4C2CA2F4.8080501@2ndquadrant.com>
Date: Thu, 01 Jul 2010 15:15:16 +0100
From: Greg Smith <greg@2ndquadrant.com>
User-Agent: Thunderbird 2.0.0.24 (X11/20100317)
MIME-Version: 1.0
To: Joachim Worringen <joachim.worringen@iathh.de>
CC: postgresql performance <pgsql-performance@postgresql.org>
Subject: Re: optimal ZFS filesystem on JBOD layout
References: <4C2C94B8.6080606@iathh.de>
In-Reply-To: <4C2C94B8.6080606@iathh.de>
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 7bit

Joachim Worringen wrote:
> Potential issues that I see:
> - the ZFS ZIL will not benefit from a BBU (as it is connected to the 
> backplane, driven by the onboard-RAID), and might be too small (32GB 
> for ~2TB of data with lots of writes)?

This is a somewhat unpredictable setup.  The conservative approach would 
be to break two disks out of the larger array for the ZIL, running 
through the battery-backed cache, rather than using the SSD drives for 
that.  The best way IMHO to use SSD for PostgreSQL is to put your large 
indexes on it, so that even if the drive does the wrong thing when you 
write and the index gets corrupted you can always rebuild them rather 
than suffer data loss.  Also, index writes really benefit from being on 
something with low seek times, moreso than the ZIL or WAL.

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