Energy Efficient Multi Dimensional Host Load Aware Algorithm for Virtual Machine Placement and Optimization in Cloud Environment

Abstract

The effectiveness and elasticity of virtual machine placement has become a main concern in modern cloud computing environment. Mapping the virtual machines to the physical machines cluster is called the VM placement. In this paper we present an efficient hybrid genetic based host load aware algorithm for scheduling and optimization of virtual machines in a cluster of Physical hosts. We used two different techniques, first initial VM packing is done by checking the load of the physical host and the user constraints of the VMs. Second optimization of placed VMs is done by using a hybrid genetic algorithm based on fitness function. The presented algorithm is implemented in JAVA Net beans IDE, and Clouds simulator has been used for simulation to assess the execution and performance of our heuristics by comparison with algorithms first fit, best fit and round robin. The performance of the proposed algorithm was examined from both users and service provider’s perception. The simulation results show that our proposed algorithm uses the less number of physical servers for placing a certain number of VMs which helps to improve the resource utilization rate. The response time of our algorithm is little bit more than the first fit algorithm because of its nature of allocating VMs is based on the user constraints and past usage history of the VMs. Elevated SLA satisfaction rate and inferior load imbalance rate was observed in results. Since we used a modified version of hybrid genetic algorithm for load optimization the percentage of VM migrations had been decreased through which we can achieve the better results for load balancing along with cost reduction. The results also show that our hybrid genetic based multi dimensional host load aware and user constraints based algorithm is applicable, valuable and reliable for implementation in real data center environments.