While Percona XtraDB Cluster is great for Multimaster Replication and InnoDB, it is not good for every type of workload. Bill Karwin states MySQL's Async Replication could be better due to workload : http://dba.stackexchange.com/questions/54612/what-are-the-drawbacks-of-using-galera-cluster-instead-of-master-slave-replicati/54615#54615 If you are still looking for synchronous replication, you could look into using [DRBD][1] with [UCARP][2] or Linux Heartbeat for DB VIP and Failover Management. Under that setup, the Master (aka DRBD Primary) has a data volume mounted on DRBD, MySQL running, and a ucarp mechanism for automatic failover. The Slave (aka DRBD Secondary) does not have the data volume mount. MySQL is not running. DRBD is doing everything in terms of having a Block Device synchronized with the DRBD Primary. DRBD is best used when having failover set up withing a single data center. DRBD should never be setup to with DRBD Primary and Secondary in two different data center due to synchronization latency. I have discussed this before in these posts - `Mar 29, 2011` : http://dba.stackexchange.com/questions/1974/mysql-high-availability-failover-and-replication-with-latency/1983#1983 - `Mar 26, 2012` : http://dba.stackexchange.com/questions/15573/3-databases-best-replication-setup/15577#15577 - `Aug 20, 2012` : http://dba.stackexchange.com/questions/22788/best-solution-for-cross-datacenter-mysql-master-slave-replication/22790#22790 I recommended this for doing PostgreSQL with DRBD : http://dba.stackexchange.com/questions/14201/postgresql-failover-what-tools-should-i-use/14215#14215 The drawback in using MySQL/DRBD for High Availability is that fact there is only one instance of MySQL running. The benefit is having a full synchronized copy of the data on two disks. Even a hard failover will have InnoDB crash recovery bring back uncommitted data that is written in the double write buffer (inside ibdata1). If you want it all, use DRBD Clusters inside a data center but have DRBD Primary in one data data center using MySQL Replication in Circular Fashion with a DRBD Primary in another data center. See my first link : http://dba.stackexchange.com/questions/1974/mysql-high-availability-failover-and-replication-with-latency/1983#1983 [1]: http://www.drbd.org/ [2]: http://www.pureftpd.org/project/ucarp