Design And Implementation Of Highly Reliable Microservice System Based On Cloud Platform

In the era of rapid development of information technology,with the booming growth of the Internet economy,the business demand for microservices applications is continuously increasing.Consequently,an increasing number of microservices applications are being deployed and run on Kubernetes clusters,as Kubernetes’ mechanism is well-suited for microservices application deployment.However,deploying a large number of microservices applications in a single Kubernetes cluster can lead to performance bottlenecks and unavailability issues when the cluster experiences failures.Moreover,as the number of Internet users continues to increase,applications can face sudden high traffic requests,but Kubernetes’ elastic scaling and load balancing strategy cannot adequately distribute the traffic.Finally,Kubernetes lacks support for dynamic addition and removal of worker nodes,requiring manual operations to improve the load capacity of the cluster.To address the above issues,this thesis presents a multi-cluster container cloud platform to ensure the reliable operation of microservices applications in container clouds.Multi-cluster management and scheduling are the core functions of this container cloud platform.Firstly,this thesis designs an interface for managing and scheduling multiple clusters.Operations and maintenance personnel can use this interface to add or remove clusters based on system conditions and business needs.Additionally,microservices applications can be deployed on multiple clusters using multi-cluster deployment mode to avoid single-point failure issues.Secondly,to address the problem of Kubernetes’ elastic scaling mechanism in the face of sudden high traffic requests,this thesis designs an event-driven scaling strategy and integrates it with the KEDA component to enhance and complement Kubernetes’ elastic scaling.In terms of load balancing,to further enhance the ability of application services to handle sudden high traffic requests,this thesis designs a JSPP load balancing strategy.This strategy dynamically adjusts the weight of application replicas based on the load situation in smooth traffic conditions.After the replicas are scaled out,this strategy can prioritize requests to newly scaled replicas to enable timely sharing of user traffic,and can quickly respond to sudden high traffic requests.The container cloud platform also supports dynamic adjustment of worker nodes.By monitoring the load pressure of the cluster and detecting resource bottlenecks or idle resources,the platform triggers node addition or deletion operations.Through experimental testing,this thesis’ s multi-cluster container cloud platform exhibits good reliability,and all module functions have achieved the intended design goals,effectively solving the problems of single-cluster deployment and ensuring the reliable operation of microservices applications on the platform.