AAVP: An Innovative Autonomic Architecture for Virtual network Piloting

Abstract

In the few last years, the Internet has known an impressive success. This success has stimulated the development and the deployment of new technologies and advances applications. Nonetheless, due to its size and scope, this large network has become victim of its own success. Innovative approaches are required to overcome the shortfalls of current systems and to design Next Generation Internet. In this context, an ambitious vision of future Inernet would include network virtualization which presents a viable solution to deal with the current Internet impasse. It provides a promising way to deploy different network architectures and protocols over a shared physical infrastructure. However, in spite of its multiple advantages, network virtualization adds more complexity on network systems. A promising solution to address this huge complexity consists in developing systems which are capable of managing themselves, called autonomic computing systems or self-* systems. This paper proposes an agent-based autonomic framework which is able to self manage virtual resources. We provide a detailed description of the proposed autonomic architecture and we focus on a real test-bed implementation and testing of our framework. Experiment results show the ability of our system to self-configure its resources in order to maintain a required QoS level.