Spike testing
Spike testing is applying a sudden, sharp increase in load to a system and watching how it responds and how it recovers once the surge ends. A ticket sale opening or a link on a popular site can multiply traffic within seconds. It is a performance test and has no connection to the agile spike, a time-boxed research story.
The longer version
The word spike names two unrelated things in software. In performance testing it means a spike in traffic, the subject here. In agile teams a spike is a short investigation story that produces knowledge instead of a feature, covered under agile spike. Nothing about one carries over to the other, and a team asking for a spike in sprint planning is asking for research.
A spike test starts from the baseline, raises the load many times over in seconds, holds it briefly and drops it back. The interesting readings are the ones a gradual ramp hides. Requests may be rejected or queued, autoscaling may take minutes to bring up new instances while the first ones drown, and a downstream service may hit its rate limit. After the surge, the questions are how long latency takes to return to normal and whether the system recovers on its own or needs a restart.
Teams repeat the test with different sizes and shapes: a single jump, two close together, a ramp that is steep but not instant. Where the real events are predictable, such as a marketing email or a sale, the test copies their size. Endurance testing is the counterpart in the other direction, holding a normal load for hours instead of a surge for seconds.
Common questions
What is spike testing?
Spike testing is a performance test that applies a sudden, large rise in load to a system and observes its behaviour during the surge and its recovery afterwards.
Is a spike in testing the same as an agile spike?
Spike testing is a performance test of sudden load, while an agile spike is a time-boxed research story that a team schedules to answer a technical or design question. The two share a word and nothing else.
What does spike testing measure?
The test measures error and rejection rates during the surge, how quickly autoscaling or queues absorb the extra work, latency while the surge lasts, and the time the system takes to return to normal afterwards.
Why is spike testing different from load testing?
Load testing ramps users up to an expected level and holds it, so a system has time to adjust. Spike testing takes that time away by raising load almost instantly, as real traffic does when it arrives all at once.
When do you need spike testing?
Spike testing is needed before events that send a crowd at the same moment, such as a sale, a ticket release, a broadcast mention or a scheduled email to a large list. A product with steady, gradual traffic needs it less.
Bug reports that arrive ready to triage.
Collect replay, console, network and browser context with the Free plan.