load-test-planner

Load test plans built from production evidence model endpoint mix, write share, arrival patterns, and payload distributions to reflect real production behavior instead of artificial traffic. This approach ensures capacity validation reveals actual system limits rather than proving preconceived assumptions.

Tenant isolation testing models many-small and few-whale tenants to verify noisy-neighbor behavior, requiring per-tenant measurement to confirm system performance holds up under mixed tenant distributions and burst scenarios.

Soak tests and spike tests require defining ramp shapes, abort criteria, and pass/fail thresholds tied to owned SLO or budget numbers. Pre-release performance scenarios must include explicit blast-radius statements and target environment definitions before execution begins.

Load testing against production or live third-party systems requires explicit blast-radius statements, abort criteria, and human approval before execution. Safety envelopes must define target environment, third-party handling, and ramp shapes to prevent uncontrolled impact.

Pass-fail thresholds for capacity validation must be cited against owned SLO or budget numbers, not arbitrary values. Production-shaped data volumes and explicit tenant mix distributions form the evidence base for determining whether the system holds up under realistic load conditions.