Energy efficiency is one of the key factors impacting the green behavior and operational expenses of telecommunication core network operations. This thesis study is aimed for finding out possible technique to reduce energy consumption in telecommunication infrastructure nodes. The study concentrates on traffic management operation (e.g. media stream control, ATM adaptation) within network processors [LeJ03], categorized as control plane. The control plane of the telecommunication infrastructure node is a custom built high performance cluster which consists of multiple GPPs (General Purpose Processor) interconnected by high-speed and low-latency network. Due to application configurations in particular GPP unit and redundancy issues, energy usage is not optimal.
In this thesis, our approach is to gain elastic capacity within the control plane cluster to reduce power consumption. This scales down and wakes up certain GPP units depending on traffic load situations. For elasticity, our study moves toward the virtual machine (VM) migration technique in the control plane cluster through system virtualization. The traffic load situation triggers VM migration on demand. Virtual machine live migration brings the benefit of enhanced performance and resiliency of the control plane cluster. We compare the state-of-the-art power aware computing resource scheduling in cluster-based nodes with VM migration technique. Our research does not propose any change in data plane architecture as we are mainly concentrating on the control plane. This study shows, VM migration can be an efficient approach to significantly reduce energy consumption in control plane of cluster-based telecommunication infrastructure nodes without interrupting performance/throughput, while guaranteeing full connectivity and maximum link utilization.