For the disk resources in the multiplex, it is best to use systems with built-in protection against hardware failures. Techniques for this protection include disk mirroring, RAID (redundant array of independent devices) disk arrays, and logging file systems.
The goal is to preserve disk contents in the most recent state, in terms of completed I/O write operations for the database server, despite any failures in components of the disk subsystem. With this guarantee, you can always restart the database, because the server will perform automatic crash recovery to bring the database to a consistent state.
Without stable disk storage, failures in the disk resources will require you to restore the disk resources to an earlier, known good state. The two ways to do this are OS-level backup and database backup. The following sections discuss these techniques.
