I have replication setup as part of my Koha installs. I have MySQL on a seperate server because we have several websites and in-house webapps that use MySQL. I also have a redundant server capable if I.P. failover. I use replication to keep the two servers in sync. You have to use a master-master setup in this kind of case though, because if the redundant server becomes the active node then all writes will happen on it, so it has to be master-master to keep both in sync with each other. I have Pace Maker and Corosync handling the I.P. failover.
MySQL replication is usually used to setup up read-only copies for load-balancing in very large installations, or for partial fault tolerance. Usually it's master-slave and the slave servers are read-only. This setup with master-master and I.P. failover for full fault tolerance is very complex and can be a headache to maintain. But I can pretty much reboot either server with a little preparation when ever I need to with very little down time, so very little chance of missing a write operation.