The increasing usage of Internet services and shifting towards cloud computing cause an increase interest in large, geographically distributed systems. Many Internet companies have been doubling the numbers of their servers, recently. These servers are grouped into several clusters in different geographic locations. The electricity used to power these distributed systems results in growing energy costs, which can not be ignored by companies. Therefore, two areas have been received substantial attention: increasing energy costs and growing carbon footprints.

Recently Power-Demand Routing (PDR for short) has been proposed in   : 1 . It consists in intelligent power demand distribution minimizing the total operating cost of a system (performance, network costs, and energy costs) by rerouting request traffic between clusters, i.e. requests from a cluster at one geographic location are rerouted to another. It is evident that in this case the system can move a portion of its electric power demand between these locations. PDR technique monitors geographic and temporal variations in energy costs and dynamically implements request routing policy by changing the load among locations in order to take into account differentials in energy costs. It is worth to pointing out that rerouting requests among Internet replicas can be done very efficiently by modifying only a small amount of information, for instance modifying routing tables. This can move megawatts of power-demand among locations. PDR can be also applied to reduce carbon footprints by moving power-demand to locations, where clean energy can be consumed.