2012 41st International Conference on Parallel Processing Workshops (2012)
Pittsburgh, PA, USA USA
Sept. 10, 2012 to Sept. 13, 2012
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/ICPPW.2012.95
Energy consumption in contemporary data centers is continuously surging in an unsustainable way, and tremendous efforts have been devoted both from industry and academia to address this issue. The pervasive utilization of virtual machines (VMs) offers the service provider an opportunity to consolidate various workloads onto fewer servers, thus improving resource utilization as well as eliminating idle power consumption. Thus the visibility of power consumption at VM granularity is especially important for the virtualized data centers to consolidate VMs effectively, which helps to improve resource utilization as well as guarantees no severe performance degradation is introduced. In this paper, we build a power model correlating VM resource usage and power consumption in a unitary perspective. The experimental results show that individual VM power consumption can be predicted with an average error of less than 5%.
Servers, Power demand, Power measurement, Virtual machining, Benchmark testing, performance profiling, cloud computing, KVM, power metering, power model, power capping
H. Yang, Q. Zhao, Z. Luan and D. Qian, "UVMPM: A Unitary Approach for VM Power Metering Based on Performance Profiling," 2012 41st International Conference on Parallel Processing Workshops(ICPPW), Pittsburgh, PA, USA USA, 2012, pp. 614-615.