Analyzing Local Strategies for Energy-Efficient Networking

Abstract

Abstract. Power management strategies that allow network infrastructures to achieve advanced functionalities with limited energy budget are expected to induce significant cost savings and positive effects on the environment, reducing Green House Gases (GHG) emissions. Power consumption can be drastically reduced on individual network elements by temporarily switching off or downclocking unloaded interfaces and line cards. At the state-of-the-art, Adaptive Link Rate (ALR) and Low Power Idle (LPI) are the most effective local-level techniques for lowering power demands during low utilization periods. In this paper, by modeling and analyzing in detail the aforementioned local strategies, we point out that the energy consumption does not depend on the data being transmitted but only depends on the interface link rate, and hence is throughput-independent. In particular, faster interfaces require lower energy per bit than slower interfaces, although, with ALR, slower interfaces require less energy per throughput than faster interfaces. We also note that for current technologies the energy/bit is the same both at 1 Gbps and 10 Gbps, meaning that the increase in the link rate has not been compensated at the same pace by a decrease in the energy consumption.

An experimental power profile of Energy Efficient Ethernet switches

“…This is far from the proportional relation and results in poor energy efficiency as networks tend to be lightly loaded [26]. Similar results are obtained for commercial Ethernet switches in [23], and in our study of an FPGA-based prototype router in [25]. We will corroborate this result for small Ethernet switches in our experimental study by considering an 8-port Cisco Catalyst switch that does not have EEE support.…”

“…The power consumption of switches and routers has been characterized in different studies such as [23,24] and our own study in [25]. The results show that once the switch is powered on and its ports are activated, the power consumption is close to its maximum value.…”

Joint overlay routing and relay assignment for green networks

“…We consider a system consisting of network switches, some of which can be put into sleep mode. Moreover, the links between switches are Ethernet cables and as in [41], they can be put into sleep mode if both ends of the link can be put into sleep mode.…”

A hybrid load-balancing and energy-aware RWA algorithm for telecommunication networks

“…Accordingly, in [12,13] the reduction of both energy costs and GHG emission through use of renewable energy sources has been investigated. To achieve the needed optimizations new routing algorithms are envisioned, taking into consideration the availability of low-power or energy-proportional operating modes [14] as well as the type of power feed given to the wavelength switches (green, dirty), according to the information provided by the Smart electric grids [15]. Also in [16] a dynamic routing approach, with paths that are established on-demand between nodes powered with renewable energy, is proposed.…”