Design And Implementation Of Container Cloud Service Management System Based On Multi-Type Service Hybrid Scaling Technology

To solve the manageability and efficiency of distributed resources,more and more enterprises choose to deploy services to the cloud as containers,set up a container-based cloud service management system,and combine auto-scaling technology to cope with dynamic service workload,to achieve highly automated operation and maintenance.However,with the growing scale of services,traditional cloud platforms often have limited scalability in the face of highly dynamic workload of multi-type services.Especially,there is often a delay in the scaling action of stateful services,causing waste of resources or degradation of service quality.Therefore,in the face of dynamic workload of multi-type services,how to improve the utilization of system resources and reduce the operation and maintenance costs while guaranteeing high-quality services for users through automatic scaling technology is a challenging issue.This paper focuses on the hybrid scaling problem of multi-type services,designs and implements a container-based cloud service management system based on the hybrid scaling technology of multi-type services.First,this paper investigates and analyzes the architecture design and autoscaling technologies of the current mainstream cloud platforms,designs and implements the container-based cloud service management system,modularizes and encapsulates the underlying virtual resource management methods,implements the unified module management interfaces,supports the simple management of the container cloud service life cycle,and provides visual maintenance tools for service building,publishing,running,and governing.Secondly,based on deep reinforcement learning and time series prediction technology,a hybrid scaling method for multi-type services is presented,which combines online learning with anomaly correction mechanism.This method uses workload prediction and deep reinforcement learning models to make online scaling decisions in advance,considering the impact of scaling time of multi-type services,and ensures the timeliness of scaling by combining scaling in both vertical and horizontal directions.At the same time,this method combines an anomaly correction mechanism,and makes a correction decision according to the real-time environment for anomalous workloads,which ensures the accuracy of scaling.Finally,the evaluation shows that the system designed and implemented in this paper meets the management requirements of container-based cloud services and can automatically scale resources timely and accurately in the face of dynamic changes of multi-type service workloads,avoiding waste of resources and degradation of service quality.