Stress testing

Stress testing is a performance testing technique that subjects a system to load beyond its designed capacity in order to find where, and how, it breaks. Where load testing progressively raises demand to establish the maximum level the system can sustain while still meeting its performance requirements, stress testing pushes past that ceiling deliberately, holding the system under extreme conditions to expose its failure modes.

The questions stress testing answers are different from those of load testing. Load testing asks "how much can it handle?". Stress testing asks "what happens when it can handle no more?". The aim is not to validate normal behavior but to characterize the system at and beyond its limits: which component fails first, how the failure propagates, whether the system degrades gracefully or catastrophically, and how long it takes to recover once the load is removed.

Stress conditions can take several forms. The most common is sustained overload: driving concurrent users, request rates, or data volumes beyond the system’s rated limits and holding them there. Spike testing applies an abrupt surge rather than a gradual ramp, exposing how the system copes with sudden traffic bursts. Soak testing, or endurance testing, holds load at or near capacity for an extended period, surfacing slow-onset problems such as memory leaks, resource exhaustion, or data growth that a short run would miss.

The recovery phase is as much a part of the result as the failure itself. A system that breaks but recovers cleanly, within its availability targets, may be acceptable; one that breaks and cannot recover without manual intervention is not. For this reason stress testing overlaps with recovery testing, which specifically evaluates how a system returns to a healthy state after a failure.

Stress testing is sometimes confused with chaos engineering. The two are complementary but distinct. Stress testing probes the system’s response to excessive load; chaos engineering probes its response to failure of its components. A stress test that survives confirms the system handles overload; a chaos experiment that holds its steady state confirms it handles failure. Together with load testing they describe a system’s resilience more completely than any one of them alone.

The results of stress testing inform decisions about scalability, auto-scaling policy, rate limiting thresholds, and the headroom a service must hold in reserve. They also expose whether protective mechanisms such as circuit breakers, bulkheads and graceful degradation actually engage when they should, or whether they exist only on paper.

See also

• Performance testing
• Load testing
• Capacity
• Recovery testing
• Chaos engineering
• Resilience