Continuous benchmarking of serverless cloud providers 3

Abstract

Serverless computing has transformed the way cloud applications are developed, offering a flexible, event-driven, and pay-as-you-go model that streamlines deployment while eliminating the need for developers to manage servers. Although its adoption continues to expand, there is a notable gap in detailed performance assessments of several major cloud providers beyond widely researched platforms like AWS Lambda, Azure Functions, Google Cloud Run. This project titled "Continuous Benchmarking of Serverless Cloud Providers," aims to create a robust framework for systematically evaluating serverless functions on Alibaba Cloud Function Compute, Huawei Cloud FunctionGraph, and IBM Cloud Code Engine Functions. The study emphasizes critical metrics such as cold and warm start latencies, the influence of language runtimes, and the impact of container image sizes. The findings reveal considerable differences in performance among providers, particularly in cold and warm start latencies, where IBM Cloud has exhibited the most stable results. This report outlines the project's objectives, methodology, technical architecture, and key findings, offering valuable insights to help developers choose the right serverless platform for their needs.