A Self-Adaptive Architecture with Energy Management in Virtualized Environments

Energy management has become an important concern in virtualized environments; Cloud providers are developing and using techniques to minimize the energy consumption of their data centres. Providers rely immediately on elastic computing to scale-up or -down infrastructure resources based on incoming workloads and user settings. However, elasticity comes at a price, as adding resources may significantly increase energy usages and costs. We believe that, before scaling-up, it is possible to cope with different workloads and energy expenditure by automatically reconfiguring the software architecture of the Cloud. This paper presents a self-adaptive architecture and a set of techniques that use information about energy consumption to enforce policies to reduce cost and energy. We experiment our approach by reconfiguring different HTTP servers and web applications of the back-end servers based on the workload and energy consumption without interrupting current running services. Our approach proves more efficient when compared to elasticity methods under the same workloads.