Research On Dynamic Virtual Machine Consolidation Algorithm In Cloud Computing Environment

Cloud computing has added new impetus to the development of the era of big data.It uses virtualization technology to abstract computing resources,network resources and storage resources of cloud data centers into shared resource pools,and share them with users around the world by internet.To meet the needs of users for different resources,the cloud data center runs a large number of physical hosts,but this will generate huge energy consumption.At present,the main way to save energy in cloud data centers is virtual machine consolidation,which can improve the energy efficiency and resource utilization of cloud data centers.However,an aggressive virtual machine consolidation approach can cause physical hosts overloaded and large-scale virtual machine migration,these phenomena can lead to a decline in quality of service(QoS)of cloud computing.Therefore,how to improve energy efficiency,resource utilization and QoS by virtual machine consolidation with a small number of virtual machine migrations becomes a major challenge.In order to solve this problem,this paper analyzes the overloaded physical host detection stage,virtual machine selection stage,virtual machine placement stage in the virtual machine consolidation process,and proposes the corresponding algorithm:(1)For the problem of overloaded physical host detection in cloud data center,this paper proposes an overloaded host detection algorithm based on differential integrated moving average autoregressive model(AROD).It fully considers the strong correlation between cloud computing workloads,and uses the ARIMA model to estimate the possibility of future overload of physical hosts and migrate redundant virtual machines before physical hosts are overloaded.(2)Aiming at the problem of low efficiency and excessive migration of virtual machines in the existing virtual machine selection algorithm,this paper proposes a Virtual Machine Selection Algorithm for Enhancing Relative Migration Benefits(ERMB)to improve the relative migration benefits.This work first analyzes the change of the overall cloud data center computing resource loss from the virtual machine migration,and proposes the relative migration gain model of the virtual machine migration.Then,the model classifies invalid virtual machine migration and effective virtual machine migration,and evaluates the effectiveness of each virtual machine migration.Finally,the ERMB algorithm prioritizes efficient virtual machine migration with higher relative migration benefits,avoiding the occurrence of invalid virtual machine migration.The experimental results show that the ERMB algorithm can improve the QoS and energy efficiency of the cloud data center while reducing the number of virtual machine migrations.(3)Aiming at the problem that the physical host is prone to secondary overload after the virtual machine placement,this paper proposes a Virtual Machine Placement Algorithm for Computing Resource Reservation Constraints(CRRC).Firstly,based on the historical resource usage of physical hosts,an adaptive computing resource constraint is proposed.Then,according to the computing resource constraints and resource utilization of the physical host,a candidate physical host list that can receive the virtual machine is established and the BFD algorithm is improved.Finally,the virtual machine to be migrated is placed into the candidate physical host through the improved BFD algorithm.The experimental results show that the CRRC algorithm can improve the QoS and energy efficiency of the cloud data center under the premise of preventing secondary overload of the physical host.Finally,the above algorithm is combined into the dynamic virtual machine consolidation algorithm EQVC,and compared with the existing virtual machine consolidation algorithm under two different real-world workload trace.The experimental results show that EQVC enables a cloud data center to provide users with a high level of QoS in a low-energy state under a small number of virtual machine migrations.