Energy-efficient anycast scheduling and resource allocation in optical grids

Abstract

In recent years, optical grid networks has been used as an ideal infrastructure to support high-performance computing environment, data intensive applications and interconnection of data centers. Due to rapid increase in the high-bandwidth applications, the power consumption of communications equipment for such networks has been increasing steadily over the past decade. Therefore, energy efficient routing schemes and traffic models can be developed to reduce the energy consumption. In many applications it is possible to select the destination node from a set of possible destinations, which have the required computing/storage resources. This is known as anycasting compared to unicasting where there is only one destination for each communication. In this paper we adopt the sliding scheduled traffic model, where setup and tear down times may vary within larger window frame. We propose a novel problem that exploits knowledge of demand holding times using anycasting model. We show how the flexibility of anycast routing can lead to additional energy saving. The problem was formulated as an integer linear program to optimally schedule demands (in time) and route them in order to minimize overall network energy consumption. The problem of energy consumption is addressed by switching off idle network components in low utilization periods. We analyze the performance of the proposed approach for anycast and unicast routing models. Our simulation results demonstrate that the proposed approach can lead to significant reductions in energy consumption, compared to traditional routing schemes.