Research On Energy-aware Virtual Machine Consolidation Technology In Cloud Computing Environment

As the e-commerce,Internet of Things,and big data industries are in the ascendant,cloud computing is playing an increasingly important role.The rapid development of cloud computing has promoted the continuous expansion of the number and scale of cloud data centers,which increased the difficulty of data center management.The servers with low utilization are likely to cause a waste of resources,and the servers with high utilization will violate the Service Level Agreement(SLA).At present,energy and climate issues are becoming increasingly prominent,and it is urgent to solve the prominent high energy consumption issues in data centers.Virtual machine consolidation is a common method to reduce energy consumption in data centers.On the one hand,it reduces the number of physical machines by migrating virtual machines to reduce energy consumption.On the other hand,excessive consolidation can lead to violations of SLA.However,the existing virtual machine consolidation technology usually only pays attention to a certain aspect of energy consumption,SLA,or the number of migrations,or although two or three of them are considered at the same time,they cannot weigh the various indicators.And there are few work considering the load predicting technology for cloud data centers,predicting the status of the host in advance and migrating the virtual machines on it.In response to the above problems,this paper considers the load prediction,while taking into account the three goals of reducing energy consumption,the number of migrations and SLA violations,and proposes a virtual machine consolidation algorithm based on the ant colony system.The specific work is summarized as follows:1.Propose a virtual machine consolidation method based on ant colony system algorithms.The migration efficiency is improved by migrating virtual machines one by one,and different heuristic information and constraint conditions are designed for underloaded hosts and overloaded hosts.We add local search rules within the population,and add pheromone communication strategy between the populations.Finally,a large number of simulation experiments are carried out on Cloud Sim,an open source cloud simulation platform commonly used in academia.The results show that the algorithm in this paper can weigh the three goals of energy consumption,migration times and SLA violations well.2.Propose a kind of host load prediction algorithm based on the extreme learning machine,which reduces the number of migrations and reduces energy consumption.This paper also optimizes the target host constraint conditions in ant colony system algorithms,reducing the probability of migration failure and reducing the energy consumption by 2.4%.The migration plan for underloaded hosts is also optimized,which reduces the number of migrations by 2.4%by reducing the occurrence of migrating virtual machines in and out on the same physical machine.3.We build a Open Stack cloud platform using five servers based on ARM,and re-develop the resource optimization component called Watcher.The prediction module is added and the scheduling algorithm proposed in this article is implemented in Watcher.Then,experiments are carried out in three scenarios with all the underloaded hosts,including both overloaded hosts and underloaded hosts,and multiple instances.Experimental results show that the consolidation algorithm proposed in this paper can schedule virtual machines appropriately,reduce the energy consumption of cloud data centers,and is of great significance for reducing carbon emissions and building green data centers.