Throttle

This project implements fixed-window rate limiters using Redis to manage request limits for users based on their subscription type. It provides three variants with varying caching mechanisms:

1. Vanilla Rate Limiter: No caching; ensures consistent data by always querying Redis.
2. Periodic Pooling + RAM Caching Rate Limiter: Combines lazy caching with periodic synchronization from Redis.
3. Subscription + RAM Caching Rate Limiter: Uses Redis Pub/Sub to proactively update local caches in real-time.

---

Assumptions

• The objective is to rate limit a service with one operational endpoint.
• User types can be one of the following: normal_user, premium_user, or admin.
• Max allowed requests for user types are not changed frequently, resulting in reasonably low configuration load.
• Any user can make a minimum of 0 requests.
• The service is under reasonably high load, with multiple replicas of the server behind a load balancer.
  • This necessitates the use of a centrally accessible database (Redis) for maintaining the Request_count instead of storing it as a Python dictionary in memory.

---

Configuration Store: Redis

Schemas

1. User_Requestcount_map

   • Tracks the live number of requests received from each user.
   • Each key (user ID) has an expiry of 1 minute.

2. User_Usertype_map

   • Maps each user ID to their corresponding user type (e.g., normal_user, premium_user, admin).

3. Usertype_Maxnumrequests_map

   • Maps each user type to its respective maximum allowed requests.

---

Flask Server

• A dummy Flask-based HTTP server is implemented with one endpoint:
  • Endpoint: /submit_job
    Decorated with a Rate_limiter_check function, which implements the rate limiter before executing the core logic.

---

Features

• Rate limiting using a fixed-window approach.
• Redis-backed live request tracking with a flexible schema.
• Modular implementation of rate limiter variants:
  1. Vanilla Rate Limiter: No caching, strict consistency.
  2. Periodic Pooling + RAM Caching: Lazy caching with periodic syncs.
  3. Subscription + RAM Caching: Real-time updates using Redis Pub/Sub.

---