Background:
Rate limiting controls request rate, but not concurrent in-flight pressure on expensive handlers.
Problem:
Under bursty traffic, high concurrency can still saturate downstream dependencies despite rate limits.
Scope:
Tasks:
Acceptance Criteria:
- Requests above configured cap are rejected predictably.
- Middleware is disabled by default unless configured.
- Load tests show improved stability under burst concurrency.
Background:
Rate limiting controls request rate, but not concurrent in-flight pressure on expensive handlers.
Problem:
Under bursty traffic, high concurrency can still saturate downstream dependencies despite rate limits.
Scope:
Tasks:
Acceptance Criteria: