Browsing by Author "Cooper, Quinn"

Now showing 1 - 1 of 1
  • Thumbnail Image
    ItemOpen Access
    Expedited Load Tests for CI/CD Microservice Applications
    (2024-09-06) Cooper, Quinn; Krishnamurthy, Diwakar; Amannejad, Yasaman; Pauchard, Yves; Drew, Steve
    In recent years, microservices have become a dominant architecture in software development, offering scalability, modularity, and agility to development processes. However, ensuring optimal performance before deployment poses a significant challenge, particularly in the fast-paced environments of Continuous Integration/Continuous Deployment (CI/CD) pipelines. Traditional performance testing methods, which rely on synthetic scenarios and lengthy testing processes, can be difficult to adopt in environments where testing needs to be both realistic and quick. To address the need for accurate and responsive testing, my work proposes and implements an innovative framework that uses real-world usage traces to identify and execute a small yet essential set of performance tests. This approach aims to seamlessly integrate with CI/CD workflows, providing developers with quick feedback on performance issues and scalability constraints that may arise from changes to one or more microservices. Through a series of empirical evaluations, I compare the effectiveness of six techniques using 12 real-world traces chosen based on their timeseries characteristics. My work aims to identify a set of performance tests that can capture historically observed system behaviour and be executed within a specified time budget. In this work, I present and test 6 different techniques for load test generation. Of the 6 techniques tested, Interval Sampling (IS) was the most reliably accurate for all traces tested. Using this technique, I was able to replicate the response time distribution of a 24-hour test on our custom testbench within just a five-minute test, achieving a mean relative percent error of only 1.12% and 2.73% at the 90th and 95th percentiles. This considerable decrease in time and resources necessary for load testing and response time modelling demonstrates the efficiency and effectiveness of my approach.