A Stochastic Virtual Machine Placement Algorithm for Energy-Efficient Cyber-Physical Cloud Systems
Yan Shi, Yi Zhang, Shuyin Tao, Xin Li, Jin Yuan Sun · 2019
With the integration of cyber-physical system and cloud computing, virtual machine (VM) placement has been of great importance to the performance of cyber-physical cloud system (CPCS). This paper proposes a stochastic VM placement algorithm that takes into account the uncertainty of resource requirements while minimizing the total energy consumption in a CPCS. Different from existing approaches that use deterministic values to represent the resource demands, the proposed stochastic VM placement algorithm models the uncertainty of resource requirements as random variables, and further formulates the uncertainty-affected VM placement problem as a stochastic optimization model. The optimization objective is to minimize the total energy consumed by all servers under the constraint of a user-specified overflow probability (i.e., the probability of demanded resources exceeding the capacity of the server). To solve the formulated stochastic optimization problem, we further propose an efficient metaheuristic algorithm to search for an energy-efficient VM placement solution that can tolerate resource requirement variations. Experimental results demonstrate that, compared with traditional deterministic VM placement algorithms, the stochastic method considering uncertain resource requirements can achieve more energy-efficient placement solutions while providing guaranteed service level.