From: Richard Huxton <dev@archonet.com>
To: Dana Hudes <dhudes@tcp-ip.info>
Subject: Re: bytea or blobs?
Date: Wed, 18 Feb 2004 16:08:22 +0000
User-Agent: KMail/1.5
Cc: Dana Hudes <dhudes@tcp-ip.info>, pgsql-sql@postgresql.org
References: <E8D7DFA4-5C92-11D8-B55E-0003938366A4@ntlworld.com>
	<200402180944.42239.dev@archonet.com>
	<Pine.LNX.4.58.0402181015520.729@screamer.tcp-ip.info>
In-Reply-To: <Pine.LNX.4.58.0402181015520.729@screamer.tcp-ip.info>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200402181608.22966.dev@archonet.com>

On Wednesday 18 February 2004 15:17, Dana Hudes wrote:
> How can one measure the result of the compression -- can I see this in
> some table or with some pgsql command?

Hmm - not so far as I know.

> At what threshold does it take place, I think its 8192?
> The nasty bit is not one picture of 100kb.
> Its 20 pictures of 5kb.

I'd have thought 5KB would trigger it - 8192 is the limit for a row (which 
TOASTing is designed to remove). There are/were some technical notes on TOAST 
when it was being built/introduced - googling might well find them.

What I'd suggest is grab some suitably representative images, base64 encode 
them and see how much it takes to store 100,000 copies of them.
-- 
  Richard Huxton
  Archonet Ltd