Rate-monotonic scheduling
Assigning fixed priorities to periodic real-time tasks by frequency: the shorter the period, the higher the priority.
Chung Laung Liu and James Layland proved in 1973 that this is optimal among all fixed-priority assignments, and gave a simple test with it: if total CPU utilisation stays below n(21/n − 1) for n tasks, every deadline is met. The bound falls towards ln 2, roughly 69 percent. Exceeding it doesn't mean the system is broken, only that you now have to do the arithmetic properly.
It fits an ordinary RTOS perfectly, since all it needs is a preemptive fixed-priority scheduler. The alternative, earliest deadline first, can reach 100 percent but has to shuffle priorities at run time. The whole analysis rests on knowing each task's worst-case execution time and on no task being blocked by a lower one — the latter being exactly what priority inversion is about.