Catalog
concept#Quality Assurance#Reliability#DevOps#Observability

Load Testing

Systematic testing of applications under defined load to assess performance, stability and scalability.

Load testing is a technique for assessing system performance by applying defined user or transaction loads.
Established
Medium

Classification

  • Medium
  • Technical
  • Architectural
  • Intermediate

Technical context

CI/CD pipeline (e.g., Jenkins, GitLab CI)APM and monitoring tools (e.g., Prometheus, Grafana)Load testing tools (e.g., k6, JMeter)

Principles & goals

Test with realistic, data-driven load profilesAlways couple measurements with representative monitoringAutomated, reproducible tests as part of the pipeline
Run
Enterprise, Domain, Team

Use cases & scenarios

Compromises

  • Misinterpretation of results without context
  • Load tests can impact production if executed incorrectly
  • Overfocusing on peak load instead of sustained performance
  • Test realistic user journeys instead of only synthetic requests
  • Automate test runs regularly and reproducibly
  • Collect and correlate monitoring and logs centrally during tests

I/O & resources

  • Load profiles or user scripts
  • Monitoring and telemetry data
  • Representative test environment
  • Measurement reports with metrics and graphs
  • Recommendations for scaling and optimization
  • Prioritized list of identified bottlenecks

Description

Load testing is a technique for assessing system performance by applying defined user or transaction loads. It measures response times, throughput and stability under expected conditions and uncovers bottlenecks and scaling limits. Tests are commonly automated and executed in staging or production-like environments to ensure realistic results.

✔

  • Early identification of bottlenecks and limits
  • Better capacity and cost planning
  • Validation of SLAs and user experience under load

✖

  • Requires realistic test data and environments
  • Potentially high infrastructure effort for large-scale tests
  • Not all production conditions can be simulated exactly

  • Throughput (requests/sec)

    Number of successfully processed requests per second under load.

  • 95th-percentile response time

    Indicates the upper response time which 95% of requests fall below.

  • Error rate under load

    Share of failed requests at a defined load level.

E-commerce load test before sales event

Simulated peak loads revealed a database bottleneck; result was a sharding and caching strategy.

API throughput optimization

Load tests identified slow endpoints which were made asynchronous and buffered, improving throughput.

Cloud auto-scaling validation

Load tests validated auto-scaling targets and prevented overprovisioning during peaks.

1

Define goals and acceptance criteria; prioritize relevant scenarios.

2

Develop load profiles and test scripts; prepare test environment.

3

Run automated tests, analyze results and initiate remediation actions.

⚠️ Technical debt & bottlenecks

  • Short-term optimizations without architectural improvements
  • Incomplete test scripts and missing maintenance
  • Lack of CI/CD automation leads to manual bottlenecks
Database I/ONetwork latencyCPU or thread contention
  • Running load tests directly in production without safeguards
  • Only-peak focus: overprovisioning for short spikes
  • Incomplete scripts that do not reflect real user behavior
  • Missing warmup phase before measuring stable performance
  • Unnoticed background jobs skew results
  • Monitoring without sufficient granularity
Knowledge of performance metrics and analysisExperience with load testing tools and scriptingFundamentals of system and network architecture
System scalabilityResponse time under loadInfrastructure and operational cost
  • • Availability of realistic test data
  • • Test environments must be production-like
  • • Budget for infrastructure and tools