A Measurement Study on Serverless Workflow Services

Jinfeng Wen, Yi Liu

Abstract: Major cloud providers increasingly roll out their serverless workflow services to orchestrate serverless functions, making it possible to construct complex applications effectively. A comprehensive study is necessary to help developers understand the pros and cons, and make better choices among these serverless workflow services. However, the characteristics of these serverless workflow services have not been systematically analyzed. To fill the knowledge gap, we conduct a comprehensive measurement study on four mainstream serverless workflow services, focusing on both features and the performance. First, we review their official documentation and extract their features from six dimensions, including programming model, state management, etc. Then, we compare their performance (i.e., the execution time of functions, execution time of workflows, orchestration overhead time of workflows) under various settings considering activity complexity and data-flow complexity of workflows, as well as function complexity of serverless functions. Our findings and implications could help developers and cloud providers improve their development efficiency and user experience.