From: Bruce Momjian <bruce@momjian.us>
Message-Id: <200611202216.kAKMGBx16139@momjian.us>
Subject: Re: [Pgcluster-general] PostgreSQL Documentation of High
In-Reply-To: <4561D534.1050502@bluegap.ch>
To: Markus Schiltknecht <markus@bluegap.ch>
Date: Mon, 20 Nov 2006 17:16:11 -0500 (EST)
CC: a.mitani@sra-europe.com, pgsql-docs@postgresql.org,
	pgcluster-general@pgfoundry.org
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset="US-ASCII"

Markus Schiltknecht wrote:
> Hi,
> 
> a.mitani@sra-europe.com wrote:
> > Current generation of PGCluster is a Shared-Nothing type of multi-master
> > and syncronous replication system.
> 
> Thank you for pointing us to yet another very common distinction in the
> clustering world: shared-nothing vs. shared-disk or even
> shared-everything. We don't touch that in the current documentation. Do
> we want or need to do so?

I feel the shared-* issue splits us up like master/slave and
multi-master splits up --- it added more confusion than clarity, because
many solutions fell in the middle.

> > I think that the feature of this type of replication system is as the
> > 'Multi-Master Replication Using Clustering' chapter of your document.
> 
> Most probably, yes. Please note that it's not *my* document :-)  Bruce
> Momjian wrote most of it, with only some hints and annoying nit-picking
> from my side.
> 
> > However, Oracle RAC is a Shared-Everything type of multi-master clustering
> > system. If it set up appropriately, most of these limitations would be
> > improved.
> 
> Shared-Everything, really? I thought they did their own distributed
> shared memory or distributed locking stuff, so it would be shared-disk.
> And together with their OCFS, they would reach shared-nothing. But I
> don't really know.

Yea, gets confusing.

> @pgsql-docs: I'd strongly vote for not mentioning Oracle if we don't
> event want to mention proprietary products for PostgreSQL. There are
> enough research or ongoing projects (even some ongoing reserch projects
> ;-)  ) to mention. PgCluster-II, GORDA, Slony-II or Postgres-R come to mind.

Good point. I mentioned Oracle RAC only because it seems to be an
industry standard, so by mentioning it, people know exactly what we are
talking about.  Is there a better way?  And people do ask for Oracle
RAC, so in a way we are telling them we don't have something similar. 
As sad as that is, it is true currently.

> > Next generation of PGCluster (I named PGCluster-II) will be a
> > Shared-Everything type of multi-master clustering system as demonstrated
> > in Toronto.
> 
> Yeah, I remember that demonstration. Do you think PGCluster-II fits
> what's described under 'Multi-Master Replication Using Clustering'? Do
> you think we should explain Shared-Nothing vs. Shared-Disk vs.
> Shared-Everything there?

pgcluster is must closer to Oracle RAC, but I haven't mentioned it
because I am unsure where it is in terms of usability and stability.
Comments?

-- 
  Bruce Momjian   bruce@momjian.us
  EnterpriseDB    http://www.enterprisedb.com

  + If your life is a hard drive, Christ can be your backup. +