Real-Time Resource Monitoring Framework in a Heterogeneous Kubernetes Cluster

Abstract

Container-based virtualization is quickly becoming the standard method for developing and deploying applications. Kubernetes is a well-known open-source project that provides a platform for containerized application orchestration. In such a container-based environment, it is critical that Kubernetes can monitor the cluster computing resource as well as the resource requirements of the running pods. The goal of this paper is to introduce a generic framework which is named as Kube-Monitor for real-time monitoring of a heterogeneous Kubernetes cluster's resources, which include traditional data center devices and raspberry pi devices, as well as monitoring pod resource requirements. Kube-Monitor will collect 15 different cluster resource metrics and update their values every five seconds. As a result, the Kubernetes operator can use the Kube-monitor framework to view a comprehensive set of continuously updated resource metrics.