Research on Kubernetes' Resource Scheduling Scheme

Currently, Google's open source container orchestration tool Kubernetes (K8s for short) has become the standard of fact for deploying containerized applications on a large scale in private, public, and hybrid cloud environments. By studying the scheduling-module of K8s source code, this paper finds that when selecting node for Pod, the module only considers the current optimal node, regardless of the use of resource costs. In order to solve this problem, this paper firstly realizes the model extraction of its scheduling module, and designs and implements the simulation experiment for the model for the first time. Secondly, a large number of papers on cloud computing resource scheduling are read. In this paper, the K8s scheduling model is improved by combining ant colony algorithm and particle swarm optimization algorithm. Finally, it is scored, and the node with the smallest objective function is selected to deploy the Pod. This paper draws on the resource scheduling model of CloudSim tool and implements resource scheduling of K8s using Java language. The experimental results show that the proposed algorithm is better than the original scheduling algorithm, which reduces the total resource cost and the maximum load of the node, and makes the task assignment more balanced.