Pro-active energy management for Wide Area Networks

We present a methodology for achieving energy savings in excess of 30% in Wide Area Networks. The approach applies a limited set of pre-calculated network topology configurations derived via a Genetic Algorithm across the day. The GA determines the minimum set of resources required in order to support a given traffic demand. Information gleaned from SNMP trap messages, triggered by the use of a link utilization threshold, determine when to switch between configurations. The threshold employs moving average smoothing and is discretely readjusted over the course of a daily cycle based on anticipated basal load variations. By exploiting MT-OSPF this approach provides a scalable and flexible means of reconfiguring an infrastructure that avoids routing discontinuities, excessive computational effort and the exchange of considerable volumes of control information.