From: Peter Eisentraut <peter_e@gmx.net>
To: david_shadovitz@xontech.com (David Shadovitz),
	pgsql-sql@postgresql.org
Subject: Re: Fetch a single record
Date: Tue, 9 Dec 2003 21:21:37 +0100
User-Agent: KMail/1.5.4
References: <8b37e2e5.0312051351.4045b4e@posting.google.com>
In-Reply-To: <8b37e2e5.0312051351.4045b4e@posting.google.com>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200312092121.37830.peter_e@gmx.net>

David Shadovitz wrote:
> I'm looking for the fastest way to fetch a single record from a
> table. I don't care what record it is.
>
> Here are two techniques I've explored:
>
> 1. Use LIMIT
> SELECT * FROM myTable LIMIT 1
>
> 2. Get a valid OID and then get the record.
> SELECT MIN(oid) AS anOID FROM myTable
> SELECT * FROM myTable WHERE oid = anOID
>
> The 1st technique is slow.  (I think PostgreSQL fetches all records
> and then returns just one.)  The 2nd is faster, but not fast enough.

This is hard to believe.  The first technique should be the fasted, but 
might suffer from a bad plan.  Please make sure that you have run 
ANALYZE, and then post EXPLAIN ANALYZE output.