Characterizing commodity serverless computing platforms

Serverless computing has become a new trending paradigm in cloud computing, allowing developers to focus on the development of core application logic and rapidly construct prototype via composition independent functions. With prosperity serverless major vendors have successively rolled out their commodity platforms. However, characteristics these platforms not been systematically studied. Measuring can help select most adequate platform develop serverless-based applications right way. To fill this knowledge gap, we present comprehensive study characterizing mainstream platforms, including AWS Lambda, Google Cloud Functions, Azure Alibaba Function Compute. Specifically, conduct both qualitative analysis quantitative analysis. In analysis, compare from three aspects (i.e., development, deployment, runtime) based official documentation taxonomy characteristics. analyze runtime performance multiple dimensions with well-designed benchmarks. First, key factors that influence startup latency applications. Second, resource efficiency different 16 representative Finally, measure difference when dealing concurrent requests explore potential causes black-box fashion. Based results derive series findings provide insightful implications for vendors.