Grant: Discovery Grants Program - Individual

With the rising scale and complexity of networks and services, there is a pressing need for scalable, efficient, and reliable management solutions. Autonomic management is envisioned to be the ultimate solution. In working towards this goal, my previous research on self-management has revealed a number of fundamental issues. First, current management systems collect only rudimentary data about network devices and flows, reporting only a few basic quality-of-service metrics. There is little information processing, or correlation among the metrics. The lack of comprehensive knowledge support and formal treatment of knowledge generation impedes the effectiveness of many network and service operations. Second, the diverse distributed applications (e.g. P2P, overlay networks, tube streaming, social networks) supported by today's Internet can interfere with vital network planning and engineering operations, such as inter-domain routing and traffic engineering. There is an inability to consolidate network and service operations, as little is understood about their interactions, resulting in critical management issues, including stability and optimality. Third, malfunctions and misconfiguration caused by today's manual (and in the future by automated) configuration and administration of networks is a primary source of service disruption, and creates susceptibility to malicious attacks. A strong sense of reliability must be introduced into the design and development of the entire management system. This includes the use of formal methodologies, for verification and evaluation of the system during design and at runtime, as well as the application of self-stabilization and fault tolerance to maintain correct operation in adversarial environments. ) This research draws on our extensive experience in automated network management and autonomics research to address these critical issues. By seeking effective and practical solutions, this work will have both immediate and far reaching impact on the management of networks, by advancing the state-of-the-art in order to enable autonomic management.