testing-performance

Purpose

This skill enables performance testing using tools like k6, Locust, and JMeter to simulate load, stress, and spike scenarios, measure metrics such as p50/p95/p99 latency, enforce SLAs, and generate flame graphs for bottleneck analysis.

When to Use

Use this skill when assessing application performance under load, identifying scalability issues, validating SLAs, or optimizing code before production. Apply it in pre-release testing, CI/CD pipelines, or when debugging high-latency problems in web services or APIs.

Key Capabilities

• Run load tests with k6 using VU (virtual users) and duration settings to simulate traffic.
• Generate reports with metrics like p50 (median), p95 (95th percentile), and p99 latency from test outputs.
• Define load profiles in Locust via Python scripts for custom user behaviors and ramp-up rates.
• Enforce SLAs by setting thresholds in JMeter and checking against results.
• Create flame graphs using integrated profiling in k6 or external tools to visualize CPU/memory usage.
• Support distributed testing across multiple machines for large-scale simulations.

Usage Patterns

To perform a basic load test, select a tool based on scenario: use k6 for quick scripts, Locust for Python-based customization, or JMeter for complex scenarios with UI elements. Always define a test script first, then run with specified profiles. For CI/CD, integrate as a step that triggers on code changes. Include parameterization for environments (e.g., staging vs. production) and always analyze results post-run. Example: Script a k6 test for an API endpoint, then scale it with Locust for user simulation.