load-testing_skill

Overview

This skill designs and runs load tests to validate performance under realistic and peak conditions, identify scalability limits, and verify SLA compliance. It helps teams find bottlenecks, quantify capacity, and produce actionable recommendations for scaling and optimization.

How this skill works

Define performance targets (p50/p95/p99, throughput, concurrent users) and model realistic user workflows and traffic patterns. Select a tool (k6, Locust, JMeter, Gatling, wrk), execute baseline, ramp, spike, and endurance tests, then analyze percentiles, error rates, throughput, and resource metrics to pinpoint failure modes. Produce remediation steps and retest after fixes.

When to use it

• Before deploying major features or a new release to production
• After performance optimizations to validate gains
• During capacity planning and scaling exercises
• To verify systems meet SLA/SLI targets under expected peak traffic
• When diagnosing intermittent failures under load

Best practices

• Test in a production-like environment with realistic data
• Design user workflows and include think time between actions
• Ramp traffic gradually and include spike and endurance scenarios
• Monitor system metrics (CPU, memory, DB connections, I/O) alongside test results
• Run tests at 2–3x expected peak load and include data variation
• Treat failed requests as first-class results and investigate root causes

Example use cases

• Load test an e-commerce API to validate checkout flow at peak shopping hours
• Verify microservice scaling behavior under concurrent user spikes
• Measure response-time percentiles and error rate after a database migration
• Run endurance tests to detect memory leaks or connection-pool exhaustion
• Compare performance impact of cache layer changes or horizontal scaling