# Fixed Window Counter
# - One counter per user per time window
# - Reset counter when window expires
# Memory: O(1) per user
# Problem: 2x burst at window boundaries

# Sliding Window Log
# - Store timestamp of every request per user
# - Count requests in last 60 seconds
# Memory: O(rate_limit) per user — up to 100 entries each
# Accurate but memory-heavy at scale

# Token Bucket
# - Each user has a bucket with max 100 tokens
# - Tokens added at fixed rate (100/min)
# - Each request consumes 1 token
# Memory: O(1) per user (just token count + last_refill timestamp)
# Allows controlled bursts

# Leaky Bucket
# - Requests enter a queue (bucket)
# - Processed at fixed rate regardless of input
# Memory: O(queue_size) per user
# Smooths traffic but adds latency