Mime-Version: 1.0 (Apple Message framework v752.3)
Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed
Message-Id: <3147EF26-2649-4FDF-BD86-2DB0AF8460FC@benjaminarai.com>
Reply-To: me@benjaminarai.com
Content-Transfer-Encoding: 7bit
From: "Benjamin Arai" <me@benjaminarai.com>
Subject: Partioning tsearch2 a table into chunks and accessing via views
Date: Fri, 24 Aug 2007 17:41:48 -0700
To: PostgreSQL <pgsql-general@postgresql.org>,
 pgsql-performance@postgresql.org

Hi,

I have an application which loads millions of NEW documents each month
into a PostgreSQL tsearch2 table.  I have the initial version completed
and searching performance is great but my problem is that each time a  
new
month rolls around I have to drop all the indexes do a COPY and re-index
the entire table. This is problematic considering that each month takes
longer than the previous to rebuild the indexes and the application in
unavailable during the rebuilding process.

In order to avoid the re-indexing I was thinking of instead creating  
a new
table each month (building its indexes and etc) and accessing them all
through a view. This way I only have to index the new data each month.

Does this work?  Does a view with N tables make it N times slower for
tsearch2 queries? Is there a better solution?

Benjamin