HANDLING OVERLOAD CONDITIONS - Hard Real-Time Computing Systems: Predictable Scheduling Algorithms and Applications - page 290

J 1

J 2

J 3

0 1 2 3 4 5 6 7 8 9 10

ρ (t)

1.0

0.8

0.6

0.4

0.2

t 0.0

0 1 2 3 4 5 6 7 8 9 10

Figure 9.2 Instantaneous load as a function of time for a set of three real-time jobs.

Note that while the overload is a condition related to the processor, the overrun is a
condition related to a task (or a single job). A task overrun does not necessarily cause
an overload. However, a large unexpected overrun or a sequence of overruns can cause
very unpredictable effects on the system, if not properly handled. In the following, we
distinguish between two types of overload conditions:

Transient overload : it is an overload condition occurring for a limited duration,
in a system in which the average load is less than or equal to one ( ρ ≤ 1), but the

maximum load is greater than one ( ρ max > 1).

Permanent overload : it is an overload condition occurring for an unpredictable
duration, in a system in which the average load is higher than one ( ρ> 1).

In a real-time computing system, a transient overload can be caused by a sequence of
overruns, or by a bursty arrival of aperiodic requests, whereas a permanent overload
condition typically occurs in periodic task systems when the total processor utilization
exceeds one.