A Design and Implementation of a Self-Managed Kubernetes Mobile Edge Cluster

초록

In recent years, edge computing has acquired a significant interest in developing and deploying applications by bringing computational resources close to the data source, improving the quality of application experience, and optimizing system resource usage. Edge node deployment strategies in existing studies suppose that once edge nodes are registered in an edge computing system, they will permanently operate with high availability. However, deploying the orchestration solutions in edge computing environments is challenging because of the mobility of edge nodes (mobile nodes) and low availability due to energy limitations and network unstable. To solve this problem, we propose a standalone management structure for deploying mobile nodes in an edge-cloud computing environment that enables mobility support in a Kubernetes-based Edge-Cloud infrastructure. We develop self-managed mobile edge nodes constructed as standalone nodes, ensuring workloads continue operating even when the connection between the mobile node and the cloud is interrupted. Additionally, we consider an energy-aware deployment strategy through energy monitoring and workload scaling schemes at the self-managed mobile node. In this paper, we also implement the proposed architecture on the infrastructure built and established by OpenStack and KubeEdge open-source projects. Measurements show the feasibility of the proposed architecture in deploying mobile nodes that operate independently.

키워드