Multi-objective Scheduling Optimization Of Container Of Microservice Based On NSGA-? Algorithm

With the development of the big data,cloud computing has gradually replaced private data centers to be the most extensive application technology in enterprises.Cloud computing is a virtualized service.Users do not need to know the specific location of the service provided by the cloud computing service provider,only need to pay for the cost of their own needs to obtain the corresponding service.However,with the increasingly complex user needs of cloud computing services,single-function cloud services have been unable to meet the diverse needs of users.The emergence of hybrid clouds has solved the problem.Hybrid clouds are usually composed of private clouds and public clouds.Users can choose different clouds to allocate tasks according to their different needs.Because the cost and resource allocation of each cloud usually are different,how to choose a suitable solution for task scheduling in hybrid clouds has become an important research problem.However,the hybrid clouds contain multiple resource configurations,and many factors need to take into account.Most of the existing researches are limited to a single goal and cannot accurately describe the task scheduling scenarios in the hybrid clouds.Therefore,this paper takes into account the actual situation of cloud computing resources,and conducts multi-objective research on task scheduling problems in hybrid clouds scenarios.There are usually two modes of task scheduling in cloud computing,one is the traditional virtual machine-based scheduling strategy,and the other is the container-based scheduling strategy.Compared with the traditional virtual machine,the container has the characteristics of fast startup and light weight.After the tasks deployed in the containers are executed,the resources can be released immediately.And the next container deployed task can continue to execute.Based on the above advantages,the mode of task scheduling based on container in hybrid clouds has been widely applied.The research of this paper mainly aims at the task scheduling problem in the hybrid clouds,adopts the scheduling mode based on container,and takes the cost of renting virtual machines,the delay in the execution of tasks and the reliability of tasks as optimization goals.The NSGA-? algorithm is used to compare the task scheduling mode based on container and virtual machine.The experimental results show that in the hybrid clouds scenarios,the task scheduling mode based on container is better than the task scheduling mode based on virtual machine.In addition,this paper proposes a time interval based on elastic scheduling strategy for containers in hybrid clouds.It considers the vertical scaling,horizontal scaling,and the lease time interval of virtual machine.Explore how to set a reasonable lease time interval of virtual machine for microservice tasks of different scales in the process of microservice task scheduling.Because the crossover and mutation probability of the traditional NSGA-? algorithm is fixed,it is easy to fall into the local optimal value.Therefore,this paper adopts a dynamically changing crossover and mutation probability method,which compares the individual crowding degree with the average crowding degree of the population to dynamically change the crossover and mutation probability to help the algorithm from jump out of the optimal value.Experiment with the improved NSGA-? algorithm,NSGA-? algorithm and MOEAD algorithm under the same conditions,the experimental results show that the performance of the improved NSGA-? algorithm has been significantly improved.