🔥 Live 2,847 QA engineers learning right now — Start Free Automation Roadmap →
SDET · Performance Testing

Making systems prove they scale

Performance testing is a high-value gap most testers have not closed. Here are the four test types, the metrics that matter, and the tools — interview-ready.

Updated for 2026 · ~10 min read · by Naveed Tawargeri

Most testers can automate functional flows; far fewer can answer "how would you check this holds up under load?" Performance testing is a high-value gap to close, and the concepts are simpler than they sound once you separate the four test types.

The four performance test types

Nearly every performance question maps to one of these. Know what each one answers, a real example, the pass criteria, and the tool.

Test typeWhat it answersReal-world examplePass criteriaTool
LOAD TESTDoes the system work under NORMAL expected traffic? 'Can 100 users use Swiggy at the same time?'100 users order food simultaneously on a Monday evening — normal peak load.Error rate < 1% P95 response < 500ms No crashesJMeter / k6
STRESS TESTWhere does the system BREAK? 'How many users can Swiggy handle before it falls over?'Keep increasing users 100→200→500→1000 until errors start or app crashes.No hard pass — goal is to FIND the limit. Document breaking point.JMeter / k6
SPIKE TESTWhat happens when traffic suddenly JUMPS 10x? 'What happens when Dhoni retires and everyone opens Twitter at once?'Dream11 match starts. 50 users → 500 users in 30 SECONDS. Then drops back.System survives Auto-scaling triggers Errors return to normal after spikek6 (easiest for spike)
SOAK / ENDURANCE TESTDoes the system STAY healthy over a LONG time? 'Can the payment server run for 8 hours without slowing down?'Paytm payment server running at normal load for 8 hours non-stop.Memory usage stays flat Response time doesn't degrade No connection pool exhaustionJMeter with duration

The metrics that matter

  • Response time (P95/P99) — not the average; the 95th/99th percentile, because the slow tail is what users feel.
  • Throughput — requests handled per second.
  • Error rate — percentage of failed requests under load; usually the pass/fail line.
  • Concurrent users — how many simultaneous users the system serves within its targets.

Tools to name

JMeter (the industry standard, GUI + CLI), k6 (developer-friendly, scripts in JavaScript), and Gatling are the usual answers. Mention that you run performance tests in CI against a staging environment, not production.

Round it out

Performance sits next to your API and framework skills.

API Scenario Lab → Framework Simulator

Frequently asked questions

What are the main types of performance testing?

Load (behaviour under expected traffic), stress (where it breaks), spike (sudden surges) and soak/endurance (stability over long periods). Each answers a different question.

What is the difference between load and stress testing?

Load testing checks the system works under expected traffic within targets; stress testing deliberately pushes past capacity to find the breaking point and how it fails.

Which performance tool should I learn?

JMeter is the most widely expected; k6 is popular for scriptable, developer-friendly tests. Knowing one well and being able to name the others is enough for most interviews.

What is P95 response time?

The response time that 95% of requests come in under. It captures the slow tail users actually feel, which the average hides.

Test types and metrics are durable; tool versions change — verify current JMeter/k6 features when you go deep.

Related guides & practice