5.3 Active and Passive Metapolicies

Metapolicies share most of their properties with normal policies. Consequently, there are also passive and active metapolicies.

The state space metapolicies care about is given by the whole number of policies, the supporting processes as named in chapter , and their properties.

Passive metapolicies specify the permissible state space including valid state transitions. Active metapolicies react on perturbations of the permissible state space. With their help, a system becomes compliant with passive policies again.

In some cases it is possible to achieve the management tasks with either passive or active metapolicies. A reason for an explicit distinction may not be obvious immediately. Metapolicies assist in the management of policies and their related processes. Management tasks must generate as few expenses as possible, because management on its own offers no real value. Therefore, guiding the processes with additional information causes just minimal extra effort and is the preferable solution. This additional information source is provided by passive metapolicies. For this reason an explicit distinction of active and passive metapolicies also makes sense when the management task can be achieved by either one of them. In cases where only a passive or an active metapolicy is possible, it makes explicit the fact that the active ones cause state transitions by performing management activities.

Although it is important to distinguish between passive and active metapolicies, it depends on the actual implementation whether an active or passive metapolicy can be used. If a supporting process does not consider a stated passive policy, an (additional) active metapolicy must be used to force the process to comply with the intended behaviour. In this way, active and passive metapolicies work closely together. A real example can be seen in section , where limited attributes describing the state of a policy exist. A passive metapolicy describing the valid state transition of a policy says: ``The state of an altered policy changes to modified.'' A refinement process causing the alteration of the policy acts on behalf of this passive metapolicy and marks the altered policy as modified. It acts while not leaving the permissible state space. In another system, the process which causes the alteration does not consider the stated passive policy. In this case, an active metapolicy must ensure that the state of the policy does change. This can be realised by triggering/invoking actions on the process. Whenever this is not possible, active policies must use other solutions, for instance as described in chapter .

In subsection , the differences between active policies and processes are described. Almost all of the different characteristics can be applied to metapolicies. One exception is the consideration of organisational aspects. This is an area where active metapolicies are not as different from processes as active policies are. Because, they consider organisational aspects as they manage for instance the delegation process. On the basis of the remaining differences it is clear that active metapolicies are no process descriptions. They assist and control processes, but do not specify complex tasks necessary to move policies. Consequently, active metapolicies do not aim at describing large algorithms. They just perform very specific management activities.