Design performance test scenarios

Before a launch where load is a real unknown.

Design performance tests for the system below.

1. State the scenarios worth testing: normal peak, the worst realistic spike, sustained load over hours, and the growth expected in a year.
2. For each, the concrete numbers: concurrent users, requests per second, data volume, and the mix of operations. Base these on what I gave you and mark anything you assumed.
3. Define the pass criteria as percentiles, not averages. Say which percentile matters for this system and why.
4. Name what will break first, with reasoning: connection pools, a single query, memory, a third-party rate limit, a queue.
5. Describe what to measure beyond response time: error rate, saturation, queue depth, resource use.
6. Say what not to bother testing, because it will never be the bottleneck here.

System:

Beyond unit tests. Manual QA, exploratory charters, accessibility, load, and contracts between services.