An energy-efficient frequency scaling technique for virtualised memory in clouds

Abstract

As more and more non-trivial applications have been deployed on cloud platforms, the energy consumption in cloud-based datacenters has become a critical issue that needs to be addressed. In previous studies, most of the researchers focused on improving the CPU-related energy conservation approaches while few of them take efforts on memory-related energy saving techniques. In this paper, a novel memory energy saving mechanism called frequency scaling on virtualised memory is proposed, which uses the dynamical voltage frequency scaling on memory component so as to adjust the working power of a virtualised server's memory based on the runtime characteristics of active virtual machines. In this way, the proposed mechanism can provide a fine-grained energy conservation mechanism for those virtualised servers in cloud platforms. In the experiments, we extensively investigate the effectiveness and performance of the proposed mechanism on various kinds of servers and workloads. The experimental results show that it can significantly improve the energy-efficiency of memory subsystem in virtualised servers.