Journals are written if journaling is enabled, and these do prove effective for crash recovery. However, there is no way to perform disaster recovery using journals. Provide a means to 1) archive journals to separate files that can be copied/rsynced/etc to other locations, and 2) replay them. With this capability, a physical backup could be restored from tape, then journals copied off to another location could be replayed to bring the database back up to date.