DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta;
	h=received:message-id:date:from:user-agent:mime-version:to:cc:subject:references:in-reply-to:content-type:content-transfer-encoding:sender;
	b=J/wsZnt5Gt03nw/t1gsuNuVb7QxsbbyLvh0LOXjoOImv7oAOLyYWZg20MnJGsAtsyz0WVp0gFDjBfzEBerQMTAMHWoDt1ypMIsLnVjH2GhUQS7jWOvvQTeipAdhUQ7UQZUbLLgv1xbqc0XSrPZM98Lm4lTWis6mGpqLl895PcwA=
Message-ID: <472F781F.6080200@enterprisedb.com>
Date: Mon, 05 Nov 2007 20:07:59 +0000
From: Heikki Linnakangas <heikki@enterprisedb.com>
User-Agent: Icedove 1.5.0.12 (X11/20070730)
MIME-Version: 1.0
To: Chris Hoover <revoohc@gmail.com>
CC: PGSQL Performance <pgsql-performance@postgresql.org>
Subject: Re: Which index methodology is better?-
References: <1d219a6f0711051043x569b7f51r4cee3353e6ea9f10@mail.gmail.com>
In-Reply-To: <1d219a6f0711051043x569b7f51r4cee3353e6ea9f10@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Chris Hoover wrote:
> On this table, a customer can search by customer_id, customer_l_name,
> and customer_f_name.
> 
> Is it better to create 3 indexes, or one index on the three columns?
> 
> I did some initial testing with index customer_test_idx(customer_id,
> customer_l_name, customer_f_name) and postgres would use the index for
> select * from customer where customer_f_name = 'zxy' - so the single
> index will cover the three.

Postgres can use the index in that case, but it's going to have to scan 
the whole index, which is a lot slower than looking up just the needed 
rows. If you do an EXPLAIN ANALYZE on that query, and compare it against 
"select * from customer where customer_id = 123", you'll see that it's a 
lot more expensive.

I'd recommend having separate indexes. Having just one index probably 
does take less space, but the fact that you don't have to always scan 
all of it probably outweighs that.

-- 
   Heikki Linnakangas
   EnterpriseDB   http://www.enterprisedb.com