Protect computation from transient faults, wear, manufacturing defects, and component failures. Apply parity, error-correcting codes, redundancy, checkpointing, recovery, and reliability metrics.
