Research Task Parallel Processing Based On Container In Cloud Environment

In recent years,cloud computing technology has become the infrastructure that the Internet relies on,and it has penetrated into many fields such as transportation,finance,medical care,and military.The traditional cloud computing platform uses the virtual machine to implement the resource isolation of the platform.However,because the virtual machine must include a complete operating system,the use of the virtual machine to perform tasks will cause additional consumption of platform resources and reduce the resource utilization of the platform.At the same time,in the execution of the task,the system needs to be started first,so that the total execution time of the platform is increased,and the calculation efficiency is lowered.Especially when the task concurrency is large and the single sub-task is small,the efficiency reduction will be more significant.The problem of relatively low resource utilization and computational efficiency severely limits the further development of cloud computing technology.Container technology can be used to solve this problem.The container instance can contain only the application process and its necessary environment.The extra resource overhead is much smaller than that of the virtual machine.At the same time,the container startup speed can effectively improve the platform efficiency.The purpose of this paper is to use the container to replace the virtual machine to perform computing tasks.Under the premise of preserving the original functions of the cloud platform,the resource utilization and computational efficiency of the cloud platform are further improved.The following three aspects are mainly studied:Firstly,in view of the problem that some functions of the cloud platform cannot be used after accessing the container.This paper adds the container management and container image management module on the basis of retaining the original functions of the cloud platform,and re-implements the resource monitoring module to make the cloud platform Containers that run on them can be effectively monitored and managed.Secondly,aiming at the lack of high-performance resource scheduling strategy in the container environment,this paper designs and implements a container-based task parallel processing method in the cloud environment,which can automatically use the maximum number of containers allowed by the platform resources to perform tasks.When the task is executed,the resource utilization rate is over 95%,and the calculation efficiency is increased to 190% of the original virtual machine scheduling policy when performing the encryption task of 5000 data.At the same time,for the task pool synchronization problem in this method,when multiple containers take tasks from the task pool in parallel,there must be an abnormal situation that the same task is acquired multiple times,and a task pool that can effectively solve this problem is designed.The synchronization mechanism improves the usability of the method.Finally,for the parallel computing scenario,the task execution nodes pause the calculation when acquiring the task,which leads to the problem of reducing the task execution efficiency.This paper proposes a task prefetching method in the cloud computing environment,so that the container in the platform is performing the task.At the same time,task acquisition is performed,which improves the task execution speed of the platform by 37%.This method can also be applied to scenarios where concurrent tasks are performed on a virtual machine cloud platform.In this paper,the solution of using containers instead of virtual machines to perform computing tasks greatly improves the performance of the cloud platform,reduces the resource loss of the cloud platform,and provides a new idea for the further development of cloud computing technology.