Rate Limiting
Token bucket bursts
A client may send steady traffic at 50 req/s but occasionally bursts 200 requests in one second when a batch job starts. Compare fixed window vs token bucket for this shape. Which do you pick and why?
Intermediate drill question.