A balanced virtual machine scheduling method for energy-performance trade-offs in cyber-physical cloud systems

Abstract

The cloud computing scheme promises many salient features such as on-demand resource provisioning to users, and it therefore has drawn significant attention from the cyber-physical systems (CPS). An increasing number of CPS have been deployed in cloud platforms, and to accommodate numerous CPS applications, cloud datacenters often consist of a huge number of physical computation and storage nodes, and the number is still increasing. As a result, the electricity power consumption in cloud datacenters is considerable, currently accounting for about 1.3% of the worldwide electricity. How to reduce the energy consumption of datacenters is an economically beneficial but challenging problem. Optimizing virtual machine (VM) scheduling in datacenters by live VM migration is an appealing method to save energy consumption. However, it is still a challenge to conduct VM scheduling in an energy-efficient and performance-guaranteed manner, since VM migration can suffer from severe performance degradation while saving energy. In this paper, we propose a balanced VM scheduling method to achieve trade-offs between energy and performance in cyber-physical cloud systems. Specifically, the problem is formulated via a joint optimization model, and a balanced VM scheduling method is proposed accordingly to determine which VMs and where should be migrated, aiming at both reducing energy consumption and mitigating performance degradation. Both analytical and simulation results demonstrate the effectiveness and efficiency of our method.