A New Container Scheduling Strategy Based on Multi-criteria Decision Analysis Algorithm

• 1、School of Electronic Engineering,Beijing University of Posts and Telecommunications,Beijing 100876

Abstract：Kubernetes is the current mainstream open source container orchestration and scheduling system, which is widely used in academic and industrial cloud computing and cloud native fields. With the application of algorithms that rely on GPU parallel computing in cloud platforms, the traditional Kubernetes scheduling strategy based on a single standard cannot give full play to the full computing performance of the cluster. KSFC (Kubernetes Scheduling For Computing) is a new Kubernetes container scheduling strategy. The goal is to optimize the scheduling of containers that provide computing services in the cloud computing platform and improve the resource utilization of the entire computing cluster. Different from the Kubernetes default container scheduling policy, KSFC introduces more Node selection criteria. Based on the multi-criteria decision analysis algorithm, KSFC collects all the criteria in a decision matrix, and selects the Node node with the highest score to execute the computing container. The simulation data shows that, compared with other container scheduling strategies, KSFC significantly improves the resource utilization of the entire computing cluster, optimizes the scheduling capability of the cloud computing platform, and shortens the running time of computing containers.

Keywords： Computer Technology Kubernetes container scheduling GPU multi-criteria decision analysis algorithm KSFC