As part of the the real-time monitoring system, automatic notifications are generated based on metrics that can predict hardware failures such as the increase of the read/write error rate or increased disk latency over time. Automatic notifications are also generated on storage depletion which covers almost-full states for individual disks.