Differential energy saving algorithms in a distributed router architecture

A distributed multistage software router (MSSR) is composed by several interconnected software routers running on personal computers (PCs). The MSSR architecture overcomes scalability and performance issues of single software router by providing parallel forwarding paths. Like many networking devices, a MSSR must be sized for peak traffic load, which implies energy inefficiency at low loads. Thus, we focus on energy saving schemes to improve the router energy efficiency by dynamically adapting the MSSR architecture to the currently offered load. We first introduce an optimal energy saving algorithm defined as a mixed integer linear programming (MILP) optimization model. Then, heuristic solutions, named differential algorithms are discussed. While the optimal approach provides higher energy savings, the heuristics avoid the complete MSSR reconfiguration, thus reducing forwarding delays and minimizing service interruption. The performance evaluation shows that the proposed heuristic algorithms, that gracefully modifies the internal MSSR configuration, preserve the load proportional energy demand characteristics of the optimal algorithm, with a minimal loss of efficiency, largely compensated by algorithm simplicity.