Agent Support for Resource, Network, and Enterprise Management

Abstract

The management of distributed real-time systems has traditionally posed systems designers and administrators particular challenges, not least because of the need to collate distributed and diverse sensor data, take management decisions, and effect these decisions. These problems are compounded by the heterogeneity of the components that make up the system being managed, which typically need to be modeled at different levels. Furthermore, these components commonly reside within the sphere of control of different management personnel, and for example, decisions taken to optimize the behavior of one component might adversely affect the proper functioning of another. Agent-based systems, and multi-agent systems in particular, have been heralded as potential solutions to the distributed systems management problem [1,2]. It is argued that the agent-based approach emphasizes a number of qualities [3] which are germane in the distributed systems management context: autonomy - individual agents are autonomous decision makers, responsiveness - the agents are responsive to changes in their environment, proactivity - the agents can take initiative, and social ability - the agents can cooperate with one another at the knowledge-level. Notwithstanding the benefits of the framework provided by the agent approach, its realization as a solution to any management problem still requires significant work, not least, the provision of the necessary software infrastructure and deep modeling of the domain problem itself.