Autonomic Computing: A Goal-Based Management Architecture

As systems are growing more complex their control and configuration requirements exceeds the comprehension and ability of their operators [15]. Increasingly, the services provided by large-scale distributed systems have to fulfill a variety of user-specific quality of service requirements. The complexity of modern network information systems and the higher service availability constraints have reached a level that puts them beyond our ability to deploy, manage and keep functioning correctly through traditional techniques. A novel approach, called Autonomic Computing, proposes that the systems manage themselves according to an administrator's goals. These goals, refined into low-level policies, represent the user-level and the business-level requirements and they are distributed among Autonomic Elements that compose an Autonomous System. In this paper, we present an architecture that supports the development of self-managed and self-optimized networks. We explore different approaches to model the autonomic behavior and to render network equipments capable of correlating their actions to accomplish administrator's goals. A laboratory prototype has been developed to highlight some aspects of our proposal that demonstrate the autonomic behavior of the network in fulfilling heterogeneous VoIP and Video streaming applications QoS requirements