Performance Testing Overview and Tool Recommendations in Continuous Testing

Continuous testing encompasses non‑functional testing, and performance testing evaluates a system's ability to handle concurrent users under specific software, hardware, and network constraints.

Performance testing includes concepts such as load testing, stress testing, limit testing, and capacity testing, though industry definitions often overlap.

In practice, load testing and stress testing are frequently used interchangeably, with many teams treating them as the same activity.

Within continuous testing, performance testing is integrated into DevOps pipelines, emphasizing "testing as code" practices.

Performance testing tools simulate high concurrency using three execution models: multi‑process, multi‑thread, and coroutine. The relationship among processes, threads, and coroutines is illustrated in the diagram below.

Processes allocate system resources and have high scheduling overhead, making them less suitable for performance testing; tools like LoadRunner support processes but are costly.

Threads reduce context‑switch overhead and improve concurrency; both LoadRunner and JMeter support multi‑threaded models, with JMeter being an open‑source option.

Coroutines are user‑level lightweight threads that avoid kernel‑level context switches, improving I/O performance; Locust utilizes a coroutine‑based model.

Locust, an open‑source Python‑based tool, allows defining user behavior with Python code, supports distributed load testing, and can simulate millions of concurrent users across various protocols, making it a strong recommendation over LoadRunner for containerized environments.

Desired features for container‑friendly performance testing tools include script editing and debugging on both PC and container, server‑side execution, server‑side scenario configuration (including headless mode), and CI integration.

Performance testing metrics covered include concurrent users, maximum concurrent users, throughput, response time (user and system), resource utilization (CPU, memory, I/O, network), and wait time between requests.

Three main categories of performance testing are load/stress testing (determining maximum load under performance constraints), failure‑recovery testing (assessing system robustness after faults), and endurance testing (evaluating long‑duration stability).