5.5.1 Refinement

Metapolicies in this category govern the development and the refinement of policies, thus influencing the processes or (human) development procedures involved. Table  shows example metapolicies:

Table 5.1: Example refinement metapolicies

The examples will be discussed briefly and requirements for their application and enforcement will be investigated:

• The first two metapolicies ensure consistency in the policy structure in the sense that a policy which is a child policy must have a parent policy. If it is not a child policy it is not allowed to have a parent.

  This policy is a passive metapolicy, because no explicit actions are stated and it describes the desirable state space of a policy. It is necessary that the system can distinguish between a child and a parent policy, but it causes no problems if a policy is both, as one in the middle of a policy tree, for example.

• The aim of the third metapolicy is to trigger the reviewing process if a policy is modified. This is necessary, for instance, because unwanted management operation caused by a policy which is discovered to be wrong may do harm in large parts of the managed system. Reviewing before its enforcement reduces the probability of unwanted effects. Many other reasons exist such as checking its conformance with parent policies and check whether other policies refined from the actual one are consistent with the modification.

  This example metapolicy is an active metapolicy, because an action is specified which must be executed when a perturbation of the desirable state space is noticed. (Here, it is the (re)activation of a modified policy.) The management system must provide the information whether a policy is modified or not. Different techniques for this are discussed in chapter .

• The fourth metapolicy is an active policy and is used to ensure that the child policies are consistent with the parent policy. This is ensured in combination with the third example metapolicy.