Performance Testing with JMeter

Introduction to Performance Testing

• Load - expected users (500 concurrent checkouts).
• Stress - beyond capacity until it breaks; find the ceiling.
• Endurance/Soak - long runs catch memory leaks.
• Spike/Volume/Scalability - sudden bursts, big data, horizontal scaling.

Key Metrics

• Response time (avg, p90, p99), throughput (req/sec), error rate.
• Server side: CPU, memory, GC pauses, DB slow queries.
• SLA example: p95 under 800ms at 500 users with under 1 percent errors.

JMeter: Your First Test Plan

1. Test Plan -> Thread Group: 100 users, ramp-up 60s, loop 10.
2. Add HTTP Request sampler: GET /api/books.
3. Add Listener: Summary Report + Response Time Graph.
4. Run, read aggregate report, find the slowest sampler.

Thread Groups, Assertions, Reports

• Thread Group - virtual users, ramp-up avoids a fake thundering herd.
• Assertions - fail samples on wrong status or missing text.
• Timers - realistic think-time between clicks.
• HTML dashboard - jmeter -g results.jtl -o report/ for shareable graphs.

Key Points

• Test with production-like data volume, not 10 rows.
• Ramp gradually; spikes are a separate test.
• Always correlate slow endpoints with server metrics.