Green Fault-tolerant Many-objective Task Scheduling Modeling And Optimization Solution

After years of research and development,cloud computing has become an indispensable part of the information technology field.It realizes resource integration and sharing through dynamic resource management,and allocates resources to provide services to users on demand.With the rapid development of computer technology and communication technology,virtualization technology has also brought a huge impact on human production and life.More and more enterprises,especially small and medium-sized enterprises,are deploying applications on the cloud,which not only saves money The actual space can also provide convenient computing and storage services.However,due to the randomness in the cloud environment,the unpredictability of operating modes and user needs,it brings great challenges to task scheduling.Therefore,this paper fully considers the factors that affect the scheduling in the entire cloud task scheduling process,constructs a many-objective optimized cloud task allocation scheduling model,and designs a high-dimensional multi-objective optimized scheduling algorithm to solve the proposed model.The specific work for:In order to solve the problem that the current research only pays attention to the time and cost in the scheduling process,the traditional scheduling cannot get a better task assignment result.This paper considers the task completion time,task execution cost,virtual machine load,user satisfaction,and proposes a manyobjective cloud task scheduling model,and uses a variety of classic manyobjective optimization scheduling algorithms to analyze these on the simulation experiment platform.Independent tasks for global optimization.In addition,this paper designs experiments to analyze the impact of the number of tasks and the number of algorithm iterations on the performance of the algorithm.In order to solve the problem that the failure of computing resources on the cloud may cause the task to be delivered in time,this paper constructs a green fault-tolerant scheduling model for cloud tasks that considers task completion time,task execution cost,virtual machine load,user satisfaction and energy consumption.This model considers the fault-tolerant technology of primary and backup task based on task replication.Each task has a backup task and then these two tasks are assigned to different hosts.Since the traditional high-dimensional multi-objective algorithm has greater pressure on the convergence of the solution in the early stage of the algorithm,and the diversity of the solution is low,some outstanding individuals are lost in solving the practical problem of cloud task scheduling.In order to better solve the task scheduling problem,this paper first analyzes the influence of different probability distributions on the convergence and diversity of the algorithm,and uses the Wilcoxon test and the Friedman test to analyze the different probability distributions.The experiment proves that the normal distribution is better than other distributions,and non-parametric tests are also done on the parameters of the normal distribution.Then the paper proposes a many-objective algorithm based on the angle penalty distance of the normal distribution(RVEA-NDAPD),which imposes the convergence pressure of the normal distribution probability in the environment selection.To ensure that there is a certain probability to consider the diversity,the algorithm is compared with other advanced many-objective optimization algorithms on the DTLZ test set.The experimental results show that the proposed algorithm has better performance.Then the algorithm is used to solve the models proposed in Chapters 2 and 3.The simulation results show that RVEA-NDAPD is more suitable for cloud task scheduling problems than other algorithms,and a better task allocation strategy is obtained.