A2A: An Architecture for Autonomic Management Coordination

Abstract

Abstract.A central challenge of autonomic systems is how to discover, monitor, analyze and control configuration data to assure operational integrity. Current architectures for configuration data management focus on federating repositories that are loosely synchronized, and do not offer autonomic coordination services. We present A2A, a novel autonomic peering architecture which delivers a unified and consistent view of actual element configuration for autonomic systems and managers, and provides synchronization primitives enabling policy coordination and mediation. We discuss the different synchronization semantics and protocols used by systems and managers to access and manipulate configuration data stored in a distributed Modeler. We show how dependent or conflicting policy actions can be automatically detected, correlated and brought to mediation. The A2A architecture has been partially implemented in a large prototype system that has been successfully demonstrated in security, network configuration, and active network applications.