ozerohax/testing-load

<input_requirements> <required>Load goal (throughput/latency/stability)</required> <required>Key user scenarios</required> <required>Environment constraints and limits</required> <required>User profile and scenario distribution</required> <optional>Base metrics and current baseline</optional> <optional>Allowed thresholds (SLA/SLO)</optional> <optional>Monitoring requirements (APM/logging)</optional> </input_requirements>

<test_design> <principles> <principle importance="critical">Load reflects real traffic distribution</principle> <principle importance="critical">Metrics and thresholds are agreed upfront</principle> <principle importance="high">Scenarios are isolated and repeatable</principle> <principle importance="high">Tests are split into warm-up/peak/stability</principle> <principle importance="high">Clearly separate load/stress/spike/soak tests</principle> <principle importance="high">Add think time and realistic pauses</principle> <principle importance="medium">Metric collection is agreed with the infra team</principle> </principles> </test_design>

<execution_rules> <rule importance="critical">Do not exceed environment limits without permission</rule> <rule importance="critical">Record load configuration and environment</rule> <rule importance="high">Verify stability across repeated runs</rule> <rule importance="high">Identify degradations and saturation points</rule> <rule importance="high">Verify you are testing the correct service version</rule> <rule importance="medium">Compare against baseline and record changes</rule> <rule importance="medium">Synchronize timestamps and log correlation</rule> </execution_rules>

<metrics> <required> <metric>p50/p95/p99 latency</metric> <metric>throughput (RPS)</metric> <metric>error rate</metric> <metric>resource usage (CPU/RAM/IO)</metric> <metric>latency for key endpoints</metric> </required> </metrics>

<do_not> <item importance="critical">Do not run load tests in production without permission</item> <item importance="high">Do not use real user data</item> <item importance="high">Do not change load parameters during the test without recording it</item> <item importance="high">Do not run tests without monitoring key metrics</item> </do_not>

<example_scenarios> <scenario>Linear ramp-up to peak load</scenario> <scenario>Step load with stabilization periods</scenario> <scenario>Long soak test for stability</scenario> </example_scenarios>