Optimizing Energy Requirements for Cloud Services during Virtual Machine Live Migration

Abstract

newline Cloud computing has become the new generation of Information Technology industry in newlinerecent years worldwide, and cloud computing services are playing an important role to newlinemeet various requirements of the people in their daily lives. The proliferation of cloud newlinecomputing leads to creation of large number of data centers worldwide deploying newlinethousands of computer systems. However, cloud data centers consume vast amount of newlineenergy that causes high operating costs and carbon dioxide emission. According to newlineAmazon s estimations, energy related costs at its data centers account for 42% of the total newlineoperating cost. It is crucial to make every effort to reduce energy consumption in data newlinecenters. Virtual Machine (VM) live migration using virtualization in data centers has newlinegreat potential to decrease energy consumption up to certain level of usage. VM newlinemigration helps to utilize hardware resources of hosts, but leads to extra energy overhead newlinein data centers. In recent years, many consolidation algorithms have been designed to newlineminimize the energy consumption in data centers, but consideration of energy overhead newlineproblem during VM migrations is very rare. With this, it is important to deliver services newlineas per Service Level Agreement (SLA) without any performance degradation. This thesis newlinepresents analysis, models and algorithm for distributed consolidation of VMs in Cloud newlinedata centers. The main goal was to optimize the energy efficiency of underlying systems newlineduring VM migrations with improved Quality of Service (QoS) to cloud consumers. The newlineproposed approach is an Energy Optimizing Hybrid Genetic Algorithm that is used for newlinethe optimization of VM placements to target hosts with minimum energy overhead during newlineVM migrations. Various related performance parameters have been analyzed and newlineoptimized with genetic algorithm using defined characteristics of VMs and Physical newlineMachines. The performance of the proposed algorithm has been compared with recently newlinedesigned algorithms for energy efficiency in data centers. The simulation results in newlineMATLAB have shown that the Energy Optimizing Hybrid Genetic Algorithm (EOHGA) newlineperformed well when tackling test problems of different kinds, and scaled up well when newlinethe problem size increased with best QoS.