Bayesian network-based Virtual Machines consolidation method

Abstract

Efficient Virtual Machines (VMs) consolidation, as one of the primary methods for balancing between guaranteeing Quality of Service (QoS) and saving energy, is critical for data centers. Most existing VMs consolidation methods reallocate physical resources by adopting live VM migration. Therefore, VMs consolidation can be cast into estimating the physical resource utilization in Physical Machines (PMs) and predicting the migration probability of VMs. In this paper, we develop a Bayesian network-based estimation model (BNEM) for live VM migration, allowing a comprehensive treatment of nine actual factors in real data centers. A selection criterion of VMs to be migrated and a VM placement criterion are presented. By combining three algorithms corresponding to different phases in VMs consolidation, a hybrid Bayesian network-based VMs consolidation (BN-VMC) method is proposed. We have validated our approach by conducting a performance evaluation study using CloudSim toolkit, and the trace-driven comparison experiments are also performed. The simulation results show that the method can significantly degrade energy consumption, avoid inefficient VM migrations, and improve QoS.