When laying out volumes on disks with SFW, locate data objects that depend on each other on separate volumes, on separate disks. This ensures that a single disk failure does not destroy both data and its recovery mechanism.
Enterprise server environments often have interdependent sets of data. For example, the datasets in a database, its archive logs, and its redo log all depend on each other.
If a volume holding database data fails, causing data loss, the typical practice would be the following steps:
Repair the cause of the failure (for example, replace one or more disks).
Restore the database to some baseline from a backup copy.
Play the archive and redo logs against the restored copy to bring the database state as close to current as possible.
If the database logs reside on the same volume as the data, however, database recovery is impossible, because both data and logs are inaccessible.