Date: Mon, 26 Sep 2005 14:42:53 -0400
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Announce <truthhurts@insightbb.com>
Cc: pgsql-performance@postgresql.org
Subject: Re: int2 vs int4 in Postgres
Message-ID: <20050926184253.GA8295@surnet.cl>
Mail-Followup-To: Announce <truthhurts@insightbb.com>,
	pgsql-performance@postgresql.org
References: <KBEKKNMFLELKGIADDEPEGEOKCAAA.truthhurts@insightbb.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <KBEKKNMFLELKGIADDEPEGEOKCAAA.truthhurts@insightbb.com>
User-Agent: Mutt/1.5.10i

On Mon, Sep 26, 2005 at 12:54:05PM -0500, Announce wrote:
> Is there an performance benefit to using int2 (instead of int4) in cases
> where i know i will be well within its numeric range? I want to conserve
> storage space and gain speed anywhere i can, but i know some apps simply end
> up casting 2byte data to 4byte (like Java int/short).
> 
> These int2 values will be used in primary and foreign key fields and I know
> that i must explicitly use tick marks (ex: where int2_column = '12') in
> order to make use of indexes, but my question is IS IT WORTH IT?  IS THERE
> ANY REAL GAIN FOR DOING THIS?

> An simple scenario would be:
> 
> Songs
> -------
> song_id serial pkey
> genre   int2   fkey
> title    varchar

Not in this case, because the varchar column that follows the int2
column needs 4-byte alignment, so after the int2 column there must be 2
bytes of padding.

If you had two consecutive int2 fields you would save some the space.
Or int2/bool/bool (bool has 1-byte alignment), etc.

This assumes you are in a tipical x86 environment ... in other
environments the situation may be different.

-- 
Alvaro Herrera       Valdivia, Chile   ICBM: S 39º 49' 17.7", W 73º 14' 26.8"
Voy a acabar con todos los humanos / con los humanos yo acabaré
voy a acabar con todos / con todos los humanos acabaré (Bender)