What system interruption thresholds need to be established when you develop performance metrics to indicate the impact of thresholds on uptime availability?