Files
Abstract
Due to the development of powerful single board computers, edge devices are widely adopted in real-world applications. At the same time, serverless computing receives increasing attention as a new application deployment model because of its simplicity, scalability, and event-driven processing. Therefore, there is a growing need for serverless computing on edge computing. Unfortunately, due to the lack of characterization of serverless edge computing, researchers have to conduct tedious measurements to understand the performance of applications/devices in non-systematic ways. To bridge this gap, this thesis presents EdgeFaaSBench, the first benchmarking suite for serverless on edge devices. EdgeFaasBench contains 12 different benchmark applications running on edge devices and measures various system and application-level metrics, including system utilization, application response time, cold/warm startup time, and impact of concurrency. Experiments are conducted on widely-used edge devices to show that EdgeFaaSBench collects diverse metrics to understand the effectiveness of serverless computing on edge.