Semantic Ordering in Policy-based Management

Policy-based management has emerged as a solution to the problem of resource management in an open distributed environment. However, adding any new rule to a large distributed system could lead to conflict and thus the integrity and consistency of the rule database within the system is one area where a solution is still sought. This paper describes how, in a given policy based system, to compute the full extent of all possible rules and how, by considering the semantics of actions/verbs within the system, the action set can be ordered to form a partially ordered set (poset), an ontology. Using the ontology, rules can be expanded to reveal hidden conflict at compile-time the algorithm for this expansion is provided in the paper. Without this expansion, the conflict remains hidden and only emerges at run-time. The paper also provides a mathematical method of computing the minimum set of rules that can represent any given rule set in a policy based system, i.e. a way of contracting a rule set. Finally, the paper discusses two (both static and dynamic) ways of implementing the semantic ordering in a policy based management system.