thread-index: AcfOxyr9YH0xk784T3ukSYCmdJFNcQ==
Content-Class: urn:content-classes:message
Importance: normal
Subject: Re: [pgsql-advocacy] 8.2 -> 8.3 performance numbers
From: "Simon Riggs" <simon@2ndquadrant.com>
To: "Merlin Moncure" <mmoncure@gmail.com>
Cc: "Josh Berkus" <josh@agliodbs.com>, "Jim Nasby" <decibel@decibel.org>,
	"postgresql performance list" <pgsql-performance@postgresql.org>
In-Reply-To: <b42b73150707250709g3a08af06o5d96810a6f1b82d6@mail.gmail.com>
References: <73185727-AD4A-4C97-B243-ADD2EE1875B5@decibel.org>
	<46A0EAE3.7030305@agliodbs.com>
	<1185367933.4146.13.camel@ebony.site>
	<b42b73150707250709g3a08af06o5d96810a6f1b82d6@mail.gmail.com>
Content-Type: text/plain;
	charset="iso-8859-1"
Date: Wed, 25 Jul 2007 15:22:25 +0100
Message-ID: <1185373345.4146.48.camel@ebony.site>
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit

On Wed, 2007-07-25 at 10:09 -0400, Merlin Moncure wrote:
> On 7/25/07, Simon Riggs <simon@2ndquadrant.com> wrote:
> >
> > Should you get the chance I would appreciate a comparative test for
> > TPC-E.
> >
> > 1. Normal TPC-E  versus
> > 2. TPC-E with all FKs against Fixed tables replaced with CHECK( col IN
> > (VALUES(x,x,x,...))) constraints on the referencing tables.
> >
> > I have reasonable evidence that Referential Integrity is the major
> > performance bottleneck and would like some objective evidence that this
> > is the case.
> >
> > No rush, since it will be an 8.4 thing to discuss and improve this
> > substantially in any of the ways I envisage.
> 
> just a small 'me too' here, the RI penalty seems higher than it should
> be...especially when the foreign key table is very small, and I can
> see how this would impact benchmarks.

Any measurements to back that up would be appreciated. "Turning it off"
isn't really a valid comparison because we do want to make the checks
and expect there to be some cost to that. We just want to quantify the
cost to allow prioritising our efforts to improve performance on that.

-- 
  Simon Riggs
  EnterpriseDB  http://www.enterprisedb.com