Research And Implementation Of Resource Scheduling Optimization Algorithm Based On Kubernetes

With the rapid growth of users and the increasing complexity of software systems,microservices and cloud native have become the mainstream solutions to solve this problem,among which Docker+Kubernetes have become the main technology choices for cloud native.However,the default resource scheduling algorithm provided by Kubernetes currently has some shortcomings.The default resource scheduling algorithm is more focused on scheduling efficiency,and insufficient consideration is given to the load balancing degree of the cluster after resource scheduling,resulting in a decrease in the overall resource utilization rate and service performance of the cluster.At present,the Kubernetes version does not provide a dynamic resource scheduling algorithm.Therefore,optimizing the initial resource scheduling algorithm and designing new dynamic resource scheduling algorithms are of great significance for improving the load balance and service performance of the cluster.In order to solve the above problems,this paper studies the initial resource scheduling and dynamic resource scheduling,and constructs a new initial resource scheduling model and dynamic resource scheduling model.The main research results are as follows:1.An initial resource scheduling algorithm based on improved genetic ant colony algorithm was proposed.This method improves the load balancing degree of the cluster after resource scheduling by accumulating a certain number of Pods for scheduling simultaneously.The algorithm improves the heuristic function of the original ant colony algorithm,designs a pheromone update method based on the reward and punishment mechanism,and improves the selection operator of the genetic algorithm according to the elite retention strategy and roulette algorithm.Finally,according to the advantages and disadvantages of the genetic algorithm and ant colony algorithm,an algorithm combining the genetic algorithm and the ant colony algorithm is proposed.The solution quickly searched in the early stage of the genetic algorithm is converted into the initial pheromone concentration of the ant colony algorithm,and then the ant colony algorithm is used for the final refined search.The experimental results show that the resource scheduling results of this algorithm have better performance in cluster load balancing,and also perform better in algorithm convergence speed.2.A dynamic resource scheduling algorithm based on load balancing is proposed.This method adopts a dynamic adaptive threshold that can be dynamically adjusted based on the overall load level of the cluster and the migration failure rate of Pods.The cubic exponential smoothing method is used to predict future load conditions based on the historical load data of node nodes,and then the weight of the utilization of various resource indicators in high load node nodes is used as the weight to measure the contribution of Pods,Thus,it is possible to choose the Pod to be migrated more reasonably,and finally select the target node by determining the high load type of the node and regularly maintaining a low load node queue for each resource indicator type.Through experiments,it has been proven that this method can significantly improve the load balancing degree of the cluster.