Research And Optimization Of Resource Scheduling Strategy Based On Kubernetes

The containerization technology represented by Docker has become an indispensable part of cloud computing.When there are a large number of containers to be managed,it is vital to improve the service effectiveness of the cloud platform by appropriately allocating resources to the containers to achieve load balancing.As a container orchestration tool,Kubernetes is widely used on cloud platforms.In order to further improve the load balancing of Kubernetes clusters,this article analyzes and optimizes its resource scheduling strategy.The main work content is as follows:(1)Research the default resource scheduling strategy of the Kubernetes cluster,and find out the shortcomings by analyzing its resource scheduling process and resource scheduling algorithm.In response to these problems,a monitoring module is designed on the basis of Kubernetes to optimize the overall scheduling architecture of static resource scheduling and dynamic resource scheduling.(2)Aiming at the Kubernetes default resource scheduling strategy that does not consider the overall load of the cluster during multi-Pod scheduling,an optimized static resource scheduling strategy is designed.Firstly,an optimization model for multi-Pod scheduling is established,the overall load balance of the cluster is taken as the objective function,and constraints are set to ensure that Pod can be deployed and run normally,and then a K-DPSO algorithm is designed based on the optimization model,and the particles are encoded and initialized.And iterative update of location and speed results in a static scheduling scheme that balances the cluster load.(3)Aiming at the Kubernetes default resource scheduling strategy that does not consider the overall load of the cluster during multi-Pod scheduling,an optimized static resource scheduling strategy is designed.Firstly,an optimization model for multi-Pod scheduling is established,the overall load balance of the cluster is taken as the objective function,and constraints are set to ensure that Pod can be deployed and run normally,and then a K-DPSO algorithm is designed based on the optimization model,and the particles are encoded and initialized.And iterative update of location and speed results in a static scheduling scheme that balances the cluster load.Finally,a Kubernetes cluster experimental environment was built,and the scheduling results of different resource scheduling strategies were compared to verify the effectiveness of the optimization scheme.The experimental results show that the optimized static resource scheduling strategy designed in this paper performs better load balance in the cluster when multi-Pod scheduling is performed,and the scheduling time using the K-DPSO algorithm has better performance than the standard particle swarm algorithm;the dynamic resource scheduling designed in this paper The strategy can migrate fewer Pods on high-load working nodes to low-load working nodes to achieve dynamic load balancing of the cluster;the performance of the cluster is significantly improved in terms of throughput and response time by stress testing.