Energy efficient strategy for placement of virtual machines selected from underloaded servers in compute Cloud

Abstract

Workload consolidation is a phase in Cloud datacenter where tasks are allocated among the available hosts in such a way that a minimal number of hosts is used and users’ need in terms of service level agreement (SLA) is fulfilled. To achieve workload consolidation, hosts are divided among three groups based on their utilization namely overloaded hosts, underloaded host and normal hosts. Detection of over or underloaded host is a challenging issue. Most of the existing researchers propose to use threshold values for such detection. We believe that there is a scope of improvement in existing methods of deciding underloaded hosts and subsequently taking off virtual machines (VMs) from them and placing them on other hosts. In this research, we propose Host Utilization Aware (HUA) Algorithm for underloaded host detection and placing its VMs on other hosts in a dynamic Cloud environment. We compare our proposed mechanism with existing one and with empirical analysis; it is shown that our proposal results into shutting off more number of hosts without compromising user’s workload requirement which leads to an energy-efficient workload consolidation with minimal migration costs and efficient utilization of active hosts.