Dataset & Code for: Quantifying Serverless Elasticity: The gumeter Benchmark Suite

This repository provides the complete artifact accompanying the paper “Quantifying Serverless Elasticity: The gumeter Benchmark Suite” (ICSOC 2025).The artifact contains the source code, configuration scripts, benchmark workflows, and experimental data used in the paper, enabling full reproducibility of the reported results. gumeter is a comprehensive benchmark suite designed to evaluate the elasticity of serverless computing platforms in distributed workflow execution.It supports Function-as-a-Service (FaaS) and Container-as-a-Service (CaaS) platforms, including AWS Lambda, Google Cloud Run, and IBM Code Engine, through the Lithops framework. The suite includes five representative workloads — FLOPS, Monte Carlo (Stock & Pi), Mandelbrot, and TeraSort — covering diverse scaling patterns such as steep scale-out, iterative scaling, and variable parallelism.All workloads are implemented under a fire-and-forget execution model to ensure reproducibility and minimal configuration requirements. This artifact provides: Source code and benchmark definitions. Cloud configuration templates for AWS, GCP, and IBM Cloud. Execution and deployment scripts (gumeter deploy, gumeter run-all, etc.). Experimental traces and visualization scripts reproducing the figures from the paper. Researchers can use gumeter to explore the trade-offs between elasticity, performance, and cost across heterogeneous serverless backends. GitHub repository: https://github.com/GEizaguirre/gumeterRelated publication: Quantifying Serverless Elasticity: The gumeter Benchmark Suite, ICSOC 2025.License: Apache 2.0 (or as specified in the repository).