Container Elastic Scaling Strategy Based On Kubernetes

With the booming development of big data and cloud computing,container technology plays an increasingly prominent role,Docker effectively solves the problems of low resource utilization and inconsistent software stack in traditional virtual machine technology.In a cloud computing environment,you need to manage large-scale containers,while Docker only focuses on providing containers and images themselves.Kubernetes is a leader in container cluster management systems with its powerful container orchestration capabilities.However,the elastic scaling strategy of Kubernetes is relatively single,which is often inadequate in some business scenarios.Therefore,this thesis studies the elastic scaling strategy of container cloud to improve the service quality and resource utilization of Web applications deployed on Kubernetes.Specifically,the following tasks are mainly accomplished:1.An elastic scaling strategy based on STL model decomposition is proposed to solve the problem of response delay and resource failure in Kubernetes response scaling strategy.The strategy to establish the forecast model of request sequences for Web applications,using STL model will request sequences are decomposed,the trend term of the decomposed,residual item,periodic item predicted respectively,and use the results to establish the resource allocation model based on queuing network,best Web application service replica sets,for Kubernetes scheduler according to the service implementing Web application container copy set elastic scaling.2.Based on the original Kubernetes infrastructure,the four modules of acquisition,monitoring,prediction and elastic scaling are extended,and the extension modules are designed and implemented respectively.The request acquisition module is responsible for reverse proxy Web application requests and collecting the request data.Monitoring module to obtain Web application Qo S index;The prediction module is responsible for model prediction based on the collected request sequence.The elastic scaling module is responsible for calculating the replica set of services to complete the elastic scaling.3.In order to evaluate the effect of the container cloud elastic scaling strategy based on Kubernetes,based on the extended Kubernetes platform,the accuracy of the prediction model,average response time,maximum response time,and CPU utilization of the proposed scaling strategy were verified by experiments.The results show that the error of the proposed prediction model is small and can accurately reflect the Web application request.The proposed scaling strategy can reasonably allocate resources,reduce service response time and improve service resource utilization.Research shows that the proposed STL model decomposition Kubernetes container cloud elastic scaling strategy can accurately allocate container resources according to Web application requests,which not only ensures the quality of service,but also minimizes resource waste and enriches the elastic scaling strategy of Kubernetes.