K-mMA VM selection in dynamic VM consolidation for improving energy efficiency at cloud data centre

Abstract

Dynamic virtual machine (VM) consolidation is an alternative solution for managing and optimising energy efficiency in a cloud data centre. This research proposed VM selection method in dynamic VM consolidation based on K-means clustering technique and computational model Markov normal algorithm (K-mMA). The objective of VM selection is to select proper VM, which should be migrated away from the overloaded physical machine and to avoid oversubscribe host. The VM selection method has been tested in simulation condition using CloudSim and PlanetLabs datasets with various conditions of VM instances (homogeneous and heterogeneous). The performance metrics in this research are energy consumption (EC), SLA time per active host (SLATAH), performance degradation due migration (PDM) and SLA violation (SLAV). Results experiment has shown that K-mMA could improve energy efficiency and quality of service (QOS) at cloud data centre significantly. Compared with existing method such as CFS, MMT, RC, and MC the proposed K-mMA could improve efficiency energy in cloud data centre by optimising VM selection problem up to 3.9%, 6.8%, 5.5%, and 5.3% respectively.