Ener-G: A Generic Approach for Modeling Energy Consumption

For the last couple of years, due to ecological as well as economical reasons, energy efficiency of ICT has become a major issue. Leading manufacturers of IT equipment advertise with ”green” and energy-efficient products. Steadily increasing energy consumption in data centers [1], as well as high electricity costs prompt organizations, companies, and administration to decrease the energy consumption of their IT infrastructures. With technologies like virtualization [2]–[5] and live migration of virtual machines [6], tools are given that allow a dynamic and automatic resource allocation. Combined with load-based energy consumption models of infrastructurecomponents (e.g. servers, storage, routers, switches, etc.) and the monitoring of their energy-related properties, especially their load, such techniques can be used to increase the energy efficiency of IT infrastructures. The monitoring of the load on certain infrastructurecomponents provides information whether the component is able to provide a required quality-of-service (QoS) to a service which is planned to be deployed on the component or not. The energy consumption models allow to estimate the energy consumption of the components for a specific service deployment scenario. Live migration enables the relocation of virtualized services based on the monitoring information and estimated load-dependent energy consumption of the components in a way that the overall energy consumption of the IT infrastructure is decreasing. In Section II a generic approach for modeling energy consumption is presented, using a server as an example for a component. Section III concludes this document.