While load testing checks behaviour under expected conditions, stress and spike tests explore how the system behaves at and beyond its limits. These tests reveal failure modes, recovery characteristics and whether safeguards like throttling work as intended.
Stress and Spike Test Design
Stress tests gradually increase load until the system breaks, helping you identify the maximum sustainable throughput and where bottlenecks occur. Spike tests apply sudden jumps in traffic to see how the system handles abrupt surges, such as those caused by marketing campaigns or batch jobs.
Example stress/spike profiles:
- Stress: ramp from 200 to 2000 users over 30 minutes, observe where errors rise
- Spike: jump from 100 to 800 users in 1 minute, hold for 5 minutes, then drop back
- Success criteria: system degrades gracefully, recovers without manual intervention, no data corruptionThe insights from stress and spike tests inform capacity planning, autoscaling policies and defensive design patterns.
Common Mistakes
Mistake 1 — Treating stress tests as one-off stunts
This limits learning.
❌ Wrong: Running a single extreme test and never repeating it.
✅ Correct: Incorporate stress tests periodically, especially before major traffic events.
Mistake 2 — Failing to define success criteria
This blurs results.
❌ Wrong: Saying “let’s see what happens” without clear expectations.
✅ Correct: Decide what acceptable degradation and recovery look like before running the test.