How do you respond when a monitoring system indicates a possible hardware degradation before an outage occurs?