Communication-Aware Optimization of Microservice Scheduling on Edge Computing

Edge computing is well suited to environments that send massive data over the network, such as the Internet of Things. Kubernetes is used to manage services reliably in geographically distributed edge environments. However, the default Kubernetes placement policy does not consider the edge environment, which may lead to the degradation of service. In this paper, we propose a placement and replacement method that considers the edge environment. Compared to the default placement policy, proposed method improves the service performance by reducing the data transfer delay and the number of replacements.