Research And Implementation Of Load Access Control For Kubernetes Cluster

With the development of container technology and container orchestration technology,applications are more widely deployed in Kubernetes clusters in a container manner,enabling better management and maintenance.In today’s increasingly growing network traffic,a large number of network requests occur simultaneously,causing enormous pressure on the physical resources of the cluster,affecting the stability of services within the cluster.However,Kubernetes has serious consequences for the stable operation of services in a cluster due to the simplicity and static nature of its load balancing policies and the limitations of its traffic restriction policies in terms of access traffic control.To ensure the network service quality and stability of cluster services,a dynamic load balancing model based on ant lion algorithm and a dynamic flow control service based on leaky bucket and dual rate tricolor token bucket algorithm(tr TCM)are proposed.Firstly,in terms of load balancing strategies,as the processing performance of each service node fluctuates with its load situation,this paper combines four indicators to comprehensively reflect the processing performance of the node,namely,request processing time,CPU utilization,memory utilization,and network bandwidth utilization.The optimized multi-objective ant lion algorithm(MOALO)is used to seek the optimal allocation weight,and then dynamically modify the weight value of each service node.This paper improves the MOALO algorithm: using the Tent chaotic mapping function to initialize the population individuals,enhancing the ergodic uniformity of the initial solution? Use adaptive weight functions to improve global search ability and search speed.Secondly,in terms of traffic restrictions,Kubernetes’ rich monitoring capabilities are utilized to classify the load situation of the cluster into different levels.Design and implement the request pre marking module to mark requests to access the workload with different priorities.Combining the division of load levels and the priority division of request tasks,the leaky bucket algorithm and the tr TCM algorithm have been optimized in terms of request restriction policies to achieve differentiated processing of requests with different priority levels under different load levels.Finally,using throughput,average response time,and other indicators as evaluation indicators,a comparative experiment was conducted on the load balancing strategy and flow control strategy proposed in this article with commonly used algorithms to test the performance of the optimization model.The experimental results show that the optimized dynamic load balancing model has excellent overall performance and is relatively stable in high concurrency environments.Due to its different configurations at different load levels,dynamic flow limiting services can better leverage system performance and stabilize various indicator values.Under high concurrency conditions,dynamic flow limiting differentiates processing of different requests,ensuring stable requests with high priority,and more suitable for practical application scenarios.