Myatu
20-07-2010, 10:24

This is especially useful in a load balanced environment, ie., where both servers are active and may write things to disk (the alternative to that is shared disk space like NFS, SAN, etc - but that re-introduces a single point of failure unless that too is mirrored on separate servers; more £££!).
Conflict handling (in case the same file was modified on two [or more] systems) can be automated as well, by selecting one as a "prefered" server (ie., a "master").