We had been using rdiff-backup, which was easy to do, but it makes
very little sense with database backups to bother doing that.
Instead we just rsync them over and don't bother with looking at
chunks that may or may not be duplicated. This should make the daily
backup runs much faster.
