1) It is easier to analyse and (re)design large-scale systems if they are
broken into smaller units.
2) The sub-system approach allows specialisation, where a sub-system is
only responsible for its own task and does not require information of
the objectives of the overall system.
3) Hierarchical systems allow a certain degree of fault tolerance. This is
due to the fact that if a sub-system breaks down, the overall system
does not necessarily stop working. Furthermore, due to ’module structure’
the failure is ’localised’ and hence easy to detect and and repair
12.2. Process level 141
(i.e. only the faulty module has to replaced and its connections reestablished).
The coordinator, however, is the weak point, because if
it stops working, the overall system cannot function anymore.
3) Even evolution seems to favour two-level hierachical systems. For example
in a human body the brain can be considered as being the coordinator,
whereas the rest of the body forms the sub-system level.
4) In the evolution of organisations two-level hierachical systems play a
major role. Even pre-historic tribes had a tribe leader, whom was
responsible for coordinating the actions of individual tribe members in
order to improve the overall well-being of the tribe.
In the following material it is shown how two-level hierarchical systems can
be analysed mathematically. Furthermore, it is shown how optimisation techniques
can be used to coordinate the running of a two-level hierarchical systems.
The material is a rather mathematical, but in order to understand it,
only a fairly modest background is needed in constrained optimisation theory.
This report is based on [1], which is a nice introduction into the theory
of hierarchical systems.