Function-as-a-Service on Edge for Industrial Digitalization: An Off-the-Shelf Case Study

Abstract

As Industry 4.0 gathers traction, smart manufacturers are exhibiting a significant interest in the digitalization of legacy infrastructure for realizing its potentials and benefits. Enabling digital data transmissions and processing on a platform such as cloud or edge is a common solution approach. However, due to limited capacity on the edge platforms there is a need to improve and optimize their utilization for multiple connected use cases. Building a serverless platform that develops, runs, and manages applications while avoiding infrastructure complexities on the edge is non-trivial. In this paper, we focus on enabling Functions-as-a-Service (FaaS) for edge devices using off-the-shelf solutions. We present a study on two popular IoT and FaaS-enablement platforms viz. AWS Greengrass V2 (proprietary) and OpenFaaS (open-source). We highlight their deployment mechanisms and indicate the pros and cons of each. Using an experimental setup, we also present the resource consumption (CPU and memory) for both and present the latency for executing operations as simple as the sum of two numbers and compute-intensive ones like making inferences using pre-trained machine learning models.