Improving energy efficiency using request tracing as a tool in a multi-tier virtualized environment

Abstract

Cloud computing and Virtualization has become the focus of tremendous amount of research in recent years. The advances in network bandwidth, the need for services that can be accessed anytime from anywhere and the change in ownership models are some of the factors responsible. As cloud services are intended to be ‘always on’, the energy costs to provision services are already significant with increase in both cloud services user-base and rise in global energy prices. Increase in energy consumption has a direct impact on provider's total cost of ownership and inflates the subscriber's price point. Identifying areas within datacenter hosting cloud infrastructure and services for energy inefficiencies or bottlenecks is one approach to save energy. In this paper, we propose a simulation approach to track service/application request level flow path in a virtualized multi-tier environment from initiation to completion; analyze the energy consumption of the request at every tier, identify bottlenecks that would impact the QOS criteria; and take necessary action to recover. Our simulation results show a 16% improvement in energy consumption using proposed approach when compared with unconstraint/default approach.