Research On Task Scheduling In Container Cloud Platform

With the rapid development of the Internet,the value of data has been extensively explored,and the demand for data processing by users has become more apparent.Cloud services can solve this problem quickly and easily,so cloud services have great potential for development.Traditional cloud service applications cannot meet the urgent needs of users for fast delivery and efficient resource allocation.Faced with this challenge,container technology came into being.Containers as a better resource virtualization technology will naturally be widely recognized and favored by the industry.The container cloud platform has the characteristics of small redundant environment,high resource utilization,fast deployment,and flexible expansion.The basic purpose of task scheduling of the container cloud platform is to reasonably allocate available resources according to different service needs of users to ensure that user tasks can be effectively executed.Among them,the focus of the container cloud service platform scheduling is mainly to consider how to ensure the higher quality of service(QoS)required by cloud users,and to meet the requirements of the service providers to ensure their own benefits task.In the design phase of the task scheduling model.Design a container cloud task scheduling model based on QoS target optimization.We consider that because users tend to overestimate the size of the containers they need to avoid the risk that the containers purchased due to high application requirements are not guaranteed to complete the task.The user's overestimation provides the cloud provider with an opportunity to adjust the task scheduling approach to accept new users based on the expected resource utilization rather than the number of requests.Overbooking is used as a way to serve users.This strategy refines the scheduling parameters to adjust the balance of satisfaction between the two parties.In the algorithm design stage,the collaborative optimization cultural gene task scheduling algorithm is improved.For the traditional cultural genetic algorithm,the tabu search algorithm is used as the local optimization algorithm.In the process of global optimization,it is prone to the problem that the algorithm falls into local optimum due to the unreasonable setting of crossover probability and mutation probability.This paper uses a variety of groups to control the global direction.The initialized population is directed to a variety of possible evolutionary approaches based on multiple cross mutation traits.And set up an elite strategy of co-evolution to avoid the phenomenon that the population is too large due to simultaneous operation of multiple groups,and speed up the optimization efficiency.Finally,the effectiveness of the proposed algorithm and model is verified by ContainerCloudSim.The experimental results show that the container cloud task scheduling model based on QoS target optimization can effectively improve the user's satisfaction with the service.And the collaborative optimization memetic algorithm designed by comparison with its original algorithm and common algorithm can effectively improve the convergence speed of the algorithm and the quality of the optimization result.