On the energy cost of robustness and resiliency in IP networks
Résumé
Despite the growing concern for the energy consumption of the Internet, green strategies for network and traffic management cannot undermine Quality of Service (QoS) and network survivability. In particular, two very important issues that may be affected by green networking techniques are resilience to node and link failures, and robustness to traffic variations.In this paper, we study how achieving different levels of resiliency and robustness impacts the network energy-aware efficiency. We propose novel optimization models to minimize the energy consumption of IP networks that explicitly guarantee network survivability to failures and robustness to traffic variations. Energy consumption is reduced by putting in sleep mode idle line cards and nodes according to traffic variations in different periods of the day. To guarantee network survivability we consider two different schemes, dedicated and shared protection, which assign a backup path to each traffic demand and some spare capacity on the links along the path. Robustness to traffic variations is provided by tuning the capacity margin on active links in order to accommodate load variations of different magnitude. Furthermore, we impose some inter-period constraints to guarantee network stability and preserve device lifetime. Both exact and heuristic methods are proposed.Experimentations carried out on realistic networks operated with flow-based routing protocols (like MPLS) allow us to quantitatively analyze the trade-off between energy cost and level of protection and robustness. Results show that significant savings, up to 30%, may be achieved even when both survivability and robustness are fully guaranteed, both with exact and heuristic approaches.