Research On Virtual Machine Scheduling Approach For Load Balancing In Cloud Computing

Load balancing is one of the key technologies of cloud computing,load can be evenly distributed to all nodes in the cloud computing system dynamically.Efficient load balancing algorithm plays an important role in improving user satisfaction,improving resource utilization,reducing the response time and avoiding the frequent migration of virtual machines.Based on the detailed and thorough research on the existing load balancing method for cloud computing,including load balancing algorithm based on task scheduling and based on resource scheduling,the advantages and disadvantages of various algorithm are summarized in this paper.The main objective of the paper's research is to migrate virtual machines among hosts to improve system load balancing degree(including two aspects:CPU and disk 10)while reducing the migration cost as much as possible.So,the purpose is to find out the best possible mapping scheme between hosts and virtual machines in the system.A better mapping scheme consists of two parts:the lower migration cost and the smaller standard deviation of the system load.The two parts constitute the cost factor of scheduling.Due to the local scheduling and invalid migration problem,the load balancing effects of first fit and round robin algorithm and NABM algorithm are not good.The paper proposes VM balanced scheduling algorithm,the algorithm selects the best solution from the final solution space S of the mapping scheme,which does consider the load balancing problem from a global perspective;the algorithm carries out substantive migration after obtaining the best mapping scheme,therefore,it reduces the migration cost;the algorithm uses MTALB algorithm to allocate multi-type tasks evenly to VMs,the effect of system load balancing is better.The main research work of this paper includes the following aspects:(1)Research on the virtual machine scheduling model based on genetic algorithm.This paper puts forward the concept of affinity of virtual machine,and define the calculation method of affinity index.The genetic algorithm uses the spanning tree as population encoding scheme.Driving crossover operation is to drive the affinity index of the spanning tree to increase as much as possible;individual mutation operation is to make the DH value of hosts tend to converge.Finally,the grouping selection strategy selects the individuals with the highest fitness to the next generation,so as to ensure that the fitness of the individuals in each generation population continues to increase.(2)To study the multi-type task allocation load balancing algorithm(MTALB),the algorithm evenly allocates different types of tasks to virtual machines,thus making the CPU utilization and disk 10 utilization of virtual machines similar.Then,tasks' response time is further shortened,different type resources of host can be fully utilized,the load balancing degree between the hosts has been further improved,so as to improve the efficiency of task processing and reduce the number of virtual machine migration.(3)Research on virtual machine balanced scheduling algorithm.For the initial mapping scheme S.First,we execute the MTALB algorithm to evenly allocate the tasks to the virtual machines;then,we calculate the value of vc(j,T)and vd(J,T)and use the genetic algorithm to get the last generation solution space S;then the algorithm finds the mapping scheme S with minimum cost factor in 5.Finally,we compare S with the best mapping scheme S',and use the appropriate migration model to migrate the virtual machines to the corresponding hosts according to the network congestion status.Eventually,the algorithm completes the load balancing of the hosts.Experimental results show that the algorithm proposed by this paper has overall advantages over first fit and round robin algorithm and NABM algorithm in terms of specific indicators of migration cost and system load balancing,and all the performance indicators are significantly improved.