Research on Resource Prediction Model Based on Kubernetes Container Auto-scaling Technology

Abstract

Cloud computing provides a new way for computing resource acquisition and brings about changes in software development deployment. Docker is a lightweight alternative technology of virtualization, which is simple, convenient and practical when developing and deploying applications. Kubernetes is an open source container management system based on Docker container technology. This paper studies the existing auto-scaling strategy of Kubernetes and proposes an auto-scaling optimization strategy which can solve the response delay problem in the expansion phase. This strategy uses a combination of empirical modal decomposition and ARIMA models to predict the load of Pods and adjust the number of Pods in advance according to the prediction result. Our experiment proves that the strategy can achieve the purpose of capacity expansion before the peak load and reduce the application request response time.