Detection of Fog Network Data Telemetry Using Data Plane Programming

Abstract

Fog computing has been introduced to deliver Cloud-based services to the Internet of Things (IoT) devices. It locates geographically closer to IoT devices than Cloud networks and aims at offering latency-critical computation and storage to end-user applications. To leverage Fog computing for computational offloading from end-users, it is important to optimize resources in the Fog nodes dynamically. Provisioning requires knowledge of the current network state, thus, monitoring mechanisms play a significant role to conduct resource management in the network. To keep track of the state of devices, we use P4, a data-plane programming language, to describe data-plane abstraction of Fog network devices and collect telemetry without the intervention of the control plane or adding a big amount of overhead. In this paper, we propose a software-defined architecture with a programmable data plane for data telemetry detection that can be integrated into Fog network resource management. After the implementation of detecting data telemetry based on In-Band Network Telemetry (INT) within a Mininet simulation, we show the available features and preliminary Fog resource management based on the collected data telemetry and future telemetry-based traffic engineering possibilities.