back

PROJECT

distributed rate limiter

Production-grade distributed rate limiting service implementing the Token Bucket algorithm, exposed over gRPC with idempotent retry semantics and dual-backend support.

Ensuring accurate rate limiting across distributed nodes without race conditions under extreme request loads.

Built a dual-backend system: PostgreSQL with SERIALIZABLE transactions and exponential backoff, alongside a Redis backend utilizing atomic Lua scripts for sub-millisecond execution.

view on github

skills used

  • TypeScript
  • Bun
  • gRPC
  • PostgreSQL
  • Redis
  • Docker

results

Redis backend achieved 18,500+ RPS, outperforming the Postgres implementation by 5x under load.

all projects