Date: Tue, 15 Aug 2006 09:18:20 -0400
From: Andrew Sullivan <ajs@crankycanuck.ca>
To: Jorge Godoy <jgodoy@gmail.com>
Cc: Michael Fuhr <mike@fuhr.org>,
	Andrew Sullivan <ajs@crankycanuck.ca>, pgsql-sql@postgresql.org
Subject: Re: Multiple DB join
Message-ID: <20060815131820.GA21939@phlogiston.dyndns.org>
References: <44E0EA72.2090505@indiana.edu>
	<20060815103624.GA21634@phlogiston.dyndns.org>
	<20060815113832.GA99592@winnie.fuhr.org> <87fyfy8ayo.fsf@gmail.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <87fyfy8ayo.fsf@gmail.com>
User-Agent: Mutt/1.5.9i

On Tue, Aug 15, 2006 at 09:01:03AM -0300, Jorge Godoy wrote:
> I can think that spreading processing requirements should be one.  And
> distributing load another one.  Disk space can be solved with new disks and
> tablespaces, but we can't yet distribute the load through several servers
> without partitioning.

The cost of inter-node communication isn't nothing, though.  It
strikes me as at least possible that the overhead of dblink is going
to be larger than whatever gains one makes from adding a new server. 
For only 20M rows, I find it pretty hard to believe the gain is going
to be demonstrable.

A

-- 
Andrew Sullivan  | ajs@crankycanuck.ca
Everything that happens in the world happens at some place.
		--Jane Jacobs