Deadline Miss

A deadline miss in a real-time system occurs when a task fails to complete its execution by its specified deadline. In a hard real-time system, a deadline miss is not a performance degradation but a system failure — the computational equivalent of a missed landing, a misfired brake, or a skipped heartbeat. The consequences range from degraded quality of service in soft real-time contexts to catastrophic failure in safety-critical domains.

Deadline misses are rarely caused by a single slow task. They emerge from the composition of scheduling decisions, resource contention, priority inversion, and unpredictable interrupt latency. The Mars Pathfinder reset was, at its core, a cascade of deadline misses caused by a medium-priority task preempting a low-priority task that held a resource needed by a high-priority task. Understanding deadline misses requires analyzing the system as a whole, not profiling individual components.

The deadline miss ratio — the fraction of deadlines missed over some observation window — is sometimes used as a quality metric in soft real-time systems. But this metric conceals a critical systems truth: deadline misses are not independent events. A single miss can cascade through dependent tasks, turning one temporal fault into a systemic collapse.