The PetShop Dataset - Finding Causes of Performance Issues Across Microservices.

Identifying root causes for unexpected or undesirable behavior in complexsystems is a prevalent challenge. This issue becomes especially crucial inmodern cloud applications that employ numerous microservices. Although themachine learning and systems research communities have proposed varioustechniques to tackle this problem, there is currently a lack of standardizeddatasets for quantitative benchmarking. Consequently, research groups arecompelled to create their own datasets for experimentation. This paperintroduces a dataset specifically designed for evaluating root cause analysesin microservice-based applications. The dataset encompasses latency, requests,and availability metrics emitted in 5-minute intervals from a distributedapplication. In addition to normal operation metrics, the dataset includes 68injected performance issues, which increase latency and reduce availabilitythroughout the system. We showcase how this dataset can be used to evaluate theaccuracy of a variety of methods spanning different causal and non-causalcharacterisations of the root cause analysis problem. We hope the new dataset,available at https://github.com/amazon-science/petshop-root-cause-analysis/enables further development of techniques in this important area.