Figure 3.10 The system becomes unstable. Copyright © 2005 Sun Microsystems, Inc. All rights
reserved. Reproduced by permission of Sun Microsystems, Inc.
colour (one colour for each machine) performed by different athletes do not
overlap one another. The result is not satisfactory because the temporal
constraints are not enforced for all of the activities.
In order to improve the result, we raise the gain of the scheduling algo-
rithm to the value of 0.3. The result is shown in Figure 3.9: the temporal
constraints are better enforced. In fact, the new performance value is equal
to 1.673. The drawback is that the resource constraints are not completely
enforced. If the gain value is raised over 2.073 the system becomes unstable
as depicted in Figure 3.10.
The second prototype allows users to modify the release time and due date
parameters of each task and to shift the activation and termination time
backwards and forwards. It does not allow the user to modify the duration of
the tasks. The introduction of this functionality does not affect the scheduler
architecture. It just requires the extension of the graphical user interface.
Another possible extension that would be useful is the introduction of the
concept of user session. This basically consists in the possibility of loading
and saving the problem description and the time plan from
to a stable
storage medium, e.g. the file system.
The case study exemplified the development of a tool for activities