What is Spike Testing?

What is spike testing?

Spike testing is a performance test that throws a sudden, sharp burst of traffic at a system (typically 5-10x normal load, ramped in seconds rather than minutes) and measures whether the system absorbs the spike, degrades gracefully, or fails outright. The goal is to validate elastic scaling, autoscaler reaction time, and the system's ability to recover when the spike drops.

Spike testing vs load testing

A load test ramps users up over minutes and holds at target concurrency for a sustained window — verification under expected peak. A spike test jumps from normal load to far above peak almost instantly, then drops back. The shape is the difference: load is a flat plateau, spike is a tall narrow tower.

The questions each answers differ:

• Load test: can my system meet SLOs at expected peak?
• Spike test: what happens when traffic doubles in 30 seconds? Does autoscaling kick in fast enough? Does the system recover when the spike subsides, or stay degraded?

When to run spike tests

• Before a known event with burst traffic. Product launch, viral marketing campaign, Black Friday minute-zero, scheduled push notification fanout.
• To validate autoscaling. If your infrastructure relies on AWS Auto Scaling Groups, Kubernetes HPA, or Lambda concurrent execution limits, spike tests confirm the scaler reacts faster than the spike crests.
• To test rate-limit and circuit-breaker behaviour. Does your API gateway shed load gracefully or return 5xx?
• After CDN or edge-cache changes. Does the cache absorb the spike or stampede the origin?

For the maximum sustained load a system handles (rather than the burst it can absorb), see capacity testing.

Key spike-test metrics

1. Time to stabilise. Seconds from spike start to the system returning to normal response-time percentiles. Long times indicate autoscaler latency.
2. Error rate during the spike. Brief 5xx blips are normal; sustained >1% means capacity ran out.
3. Recovery curve after spike drops. Does the system return to baseline within a minute, or stay degraded for tens of minutes (a sign of sticky failures, dead connection pools, OOM-killed workers that didn't restart)?
4. Autoscaler reaction time. Cloud-side metric: did the ASG / HPA add capacity, and when?

How to run a spike test

Same scripts as load testing, different load profile. In JMeter, use the Ultimate Thread Group plugin to define a stepped or instant ramp from a small baseline to the spike target. In k6, configure a ramping-arrival-rate scenario with a sharp ramp-up stage, hold for 60-300 seconds, then ramp back down.

Run the spike from LoadFocus when you need traffic from multiple cloud regions hitting simultaneously — single-source spike tests miss the geo-distributed pattern of a real campaign.

If you'd rather not write the spike-test scripts yourself, LoadFocus offers load testing services where engineers build and run the scenarios for your traffic profile.