Design Rate Limiting at Scale

Design a global rate limiter shared by millions of clients and services. Enforce per-user, per-tenant, and per-route limits with low latency and correct behavior under Redis or regional failures.