The engineering department introduced a standardized load testing methodology to replace inconsistent practices across teams. This approach ensures that testing is reproducible and provides actionable data for decision-making, rather than serving as an experimental check.
The methodology consists of six steps, starting with preparation. This phase requires functional, integration, and organizational readiness, including ensuring system observability through monitoring. Testing without proper monitoring is considered ineffective as it prevents the analysis of errors and performance degradation.
The second stage involves gathering requirements, which are documented in a load testing passport. Teams define Service Level Indicators (SLI) and Service Level Objectives (SLO), such as response times at the 95th percentile under a specific request rate, to establish measurable criteria.
The methodology distinguishes between methodology and test design. The latter defines specific goals, such as verifying SLO compliance or identifying system breaking points. A critical component is the load profile, which simulates real user behavior based on historical monitoring data from typical days.
The load profile must include high-intensity operations and anticipated future features to ensure relevant results. Teams validate endpoints to exclude non-testable operations, such as payment processing or authorization, which could be costly or impractical to simulate in a test environment.
