A Microservice Scheduler for Heterogeneous Resources on Edge-Cloud Computing Continuum

Although the Post-5G communication technology is expected to become an enabler for networking with numerous edge devices with ultra-low latency connection, it does not sufficiently exploit the features of cloud computing infrastructures especially from the aspect of heterogeneity in geographical locations and the underlying capability of computing resources. In this paper, we propose a method to represent the hierarchical layers and structures found in the current Edge-Cloud Computing Continuum and attempt to solve the scheduling problem of microservices on a Kubernetes-based platform. We present a novel and practical dynamic resource scheduling approach for microservices that extends the scheduling method implemented originally in Kubernetes toward geographical locations and hierarchy of resources. We evaluate the proposed optimized microservice scheduler on our prototype environment that virtualizes the underlying network topologies for heterogeneous resources. The evaluation result shows that our technique successfully decreases the end-to-end response time of tested microservice application with a three-tier architecture.