From: Simon J Mudd
Date: August 28 2008 10:11am
Subject: Thoughts on MySQL replication
List-Archive: http://lists.mysql.com/replication/1420
Message-Id: <20080828101135.GA20670@ams03.wl0.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii

Hello all,

I've been using MySQL replication for some time but come from a
replication background using Sybase.

MySQL replication works very well is quick and simple to configure but
has some limitations, notably that you can't configure a MySQL server
as a slave of  multiple masters.

One of the things which struck me is that AFAICS the MySQL replication
code has 2 major parts:
1. a part which connects to the master and pulls the binlog information
   to the local server, storing it in the relay-logs.
2. a part which applies the relay logs talking to the database as if it
   were a SQL client.

The description may not be exactly right but I think it's close.

Now if this MySQL code were put into an external program you could
configure 1 or more replication servers and solve many of the
limitations of the current code. You also simplify the MySQL server
as it's replication knowledge is reduced, and potentially can improve
replication behaviour (upgrades in the replication processes) without
affecting the MySQL server itself.

Would it be perhaps worthwhile doing something like this to allow for more
sophisticated replication configurations in larger mysql installations?

Just wonder what people think about this idea.

Regards,

Simon