SuperFlow: Performance Testing for Serverless Computing

• DOI: 10.48550/arxiv.2306.01620
• Publication Date: 2023-01-01
• Authors: Jinfeng Wen; Zhenpeng Chen; Federica Sarro; Xuanzhe Liu
• Keywords: Code (set theory)

Serverless computing is an emerging cloud paradigm that allows software engineers to develop applications as a set of functions (called serverless functions). However, accurately obtaining the performance (i.e., response latency) challenging due highly dynamic nature environment in which they run. To tackle this problem, possible solution use testing determine how many repetitions function with given input are needed cater fluctuation. end, we conduct empirical study state-of-the-art techniques for traditional on 65 collected from top-tier research venues. We find these exhibit low accuracy. Therefore, propose SuperFlow, first approach tailored specifically computing. SuperFlow incorporates accuracy check and stability obtain accurate reliable results. The evaluation demonstrates provides results 97.22% accuracy, 39.91 percentage points higher than best currently available technique. have publicly released code data facilitate future replication extension.