Users report that a database server has become extremely sluggish, though the application processes appear healthy. Running 'top' shows CPU usage is mostly idle but the '%wa' (iowait) value is consistently above 40%. You need to identify which physical device is saturated and confirm high disk latency before escalating to the storage team. Which command provides the per-device utilization and average I/O wait/service time you need?