Skip to content

LIF_Load_Testing.md

Latest commit

 

History

History
60 lines (41 loc) · 3.28 KB

LIF_Load_Testing.md

File metadata and controls

60 lines (41 loc) · 3.28 KB

LIF Load Testing

Date: 2025-09-30

Status

Accepted

Context

The project involves building and maintaining a suite of services that expose GraphQL APIs. There is a need to simulate realistic user loads in order to establish accurate performance baselines. The chosen solution must support modern protocols, including GraphQL queries and mutations, and provide readable, maintainable, and version-controllable test definitions. Additionally, the approach should be developer-friendly and complement the existing tech stack for automated testing, which includes Playwright and Typescript.

Decision

We have selected k6 as the primary tool for load and performance testing.

Justification:

  • Excellent support for REST and GraphQL APIs including posting queries with variables
  • JavaScript-based scripting fits our developer skillset and promotes readable and version-controlled tests
  • Lightweight and CLI-driven, making it easy to integrate into GitHub Actions and CI/CD pipelines
  • Active open-source project backed by Grafana Labs, with growing ecosystem and observability integration (e.g., output to Grafana, InfluxDB, etc.)
  • Built-in modular architecture with extensibility via xk6 for advanced protocols (e.g., WebSockets for GraphQL subscriptions)

Alternatives

Tool REST Support GraphQL Support Ease of Use Maintenance / Ecosystem Tradeoffs / Notes
JMeter ✅ Excellent ⚠️ Good (recent support via samplers) 🟡 Moderate (GUI-based, complex for advanced logic) ✅ Very active, large community GUI helps beginners; advanced logic is cumbersome; legacy feel
Locust ✅ Excellent 🟡 Decent (via plugins or manual scripting) ✅ High (Python-based, very flexible) ✅ Active, but plugin support varies Great for Python teams; advanced GraphQL needs effort
Gatling ✅ Strong ⚠️ Partial (no built-in GraphQL helpers) 🟡 Moderate (Scala/Java DSL) ✅ Maintained; commercial support High performance; steeper learning for non-Java teams
Artillery ✅ Good ✅ Good (supports queries, mutations, WebSockets) ✅ High (YAML + JS, quick start) 🟡 Maintained; smaller community Great for modern stacks; limited advanced analytics

Consequences

Positive:

  • Faster onboarding and adoption by developers due to JS-based scripting
  • Easy CI/CD integration improves automation and test coverage
  • Ability to reuse and version performance test cases alongside application code
  • Flexibility to extend protocol support via xk6

Negative/Trade-offs:

  • No built-in GUI for test design (may be a learning curve for QA engineers used to visual tools)
  • Large-scale distributed testing may require custom setup or paid cloud version (k6 Cloud)
  • Some advanced GraphQL features (e.g., persisted queries or subscriptions) may require manual scripting or extensions

References