Date: Mon, 7 Jan 2002 17:53:21 -0800
From: Sean Chittenden <sean@chittenden.org>
Cc: Daniel Kalchev <daniel@digsys.bg>, pgsql-hackers@postgresql.org
Subject: Re: again on index usage
Message-ID: <20020107175321.C67695@ninja1.internal>
References: <200201071841.UAA26328@dcave.digsys.bg>
	<14149.1010432467@sss.pgh.pa.us>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.2.5i
In-Reply-To: <14149.1010432467@sss.pgh.pa.us>;
	from "tgl@sss.pgh.pa.us" on Mon, Jan 07, 2002 at = 02:41:07PM

> It's difficult to tell from this what it thinks the selectivity of the
> ipdate index would be, since the rows estimate includes the effect of
> the ipaddr and router restrictions.  What do you get from just
> 
> explain
> SELECT sum(input) FROM iplog_gate200112 
> WHERE 
>   '2001-12-01 00:00:00+02' <= ipdate AND ipdate < '2001-12-02 00:00:00+02' AND 
>   '2001-12-01 00:00:00+02' <= ipdate AND ipdate < '2002-01-01 00:00:00+02';
> 

I don't know if this'd help, but if you're suming the data and running
this query often, see if a function index would help:

CREATE INDEX sum_input_fnc_idx ON iplog_gate200112 (sum(input));

-sc

-- 
Sean Chittenden