Energy-Aware Mapping and Live Migration of Virtual Networks

Abstract

Network virtualization facilitates the deployment of new protocols and applications without the need to change the core of the network. One key step in instantiating virtual networks (VNs) is the allocation of physical resources to virtual elements (routers and links), which can be then targeted for the minimization of energy consumption. However, such mappings need to support the quality-of-service requirements of applications. Indeed, the search for an optimal solution for the VN mapping problem is NP-hard, and approximated algorithms must be developed for its solution. The dynamic allocation and deallocation of VNs on a network substrate can compromise the optimality of a mapping designed to minimize energy consumption, since such allocation and deallocation can lead to the underutilization of the network substrate. To mitigate such negative effects, techniques such as live migration can be employed to rearrange already mapped VNs in order to improve network utilization, thus minimizing energy consumption. This paper introduces a set of new algorithms for the mapping of VNs on network substrates designed to reduce network energy consumption. Moreover, two new algorithms for the migration of virtual routers and links are proposed with simulation showing the efficacy of the algorithms.