Unfortunately there is nothing built into MySQL replication that
would automatically dispatch queries based on query type.
The best alternatives I can think of is MySQL Cluster, or
On Thu, Jan 20, 2005 at 09:21:09PM -0400, Jesus De Oliveira wrote:
> We're considering a cluster of four machines for a high availability and
> load balancing mysql database service, using replication to sync data
> along the cluster nodes. The client of this ha-database would be a J2EE
> app server, using container managed persistence of entity beans. The
> cluster software we're using is Linux Virtual Server, allowing the app
> server client to send his queries to a virtual IP address of the cluster,
> and dispatching each query to any of the cluster nodes.
> In this case, the developer has no control over which queries go to which
> nodes (because the nature of container managed persistence and connection
> pooling techniques), allowing for example a UPDATE send to a slave node
> (and as we know this would end up in a disaster).
> Is there a way to send all queries to the master server, and make him
> decide upon the query type if the query could be dispatched to one of the
> slaves? The main purpose is to have load balancing plus the high
> availability provided by the replication mechanism.
> The mysql cluster product satisfies this requeriment, and was evaluated
> but results really expensive due to the RAM needed to hold at least two
> replicas of our database of 20Gb.
> Jesus De Oliveira
> Consultor IT
> Baysystem Consultores C.A.
> Tlf: (0212) 339.06.96
> (0414) 300.06.46
> MySQL Replication Mailing List
> For list archives: http://lists.mysql.com/replication
> To unsubscribe: http://lists.mysql.com/replication?unsub=1
Dr. Lars Thalmann
MySQL AB, www.mysql.com