HARPY DIVISION
RUEN
Hire us

#01API gateway / Highload

Raven Gate

A high-throughput API gateway with rate limiting, JWT auth and hot route reloading.

  • Go
  • Redis
  • PostgreSQL
  • Kubernetes
  • NGINX Ingress
  • Prometheus
Source code is covered by an NDA Discuss a similar project

Project gallery

05 photos
Raven Gate — 1 Raven Gate — 2 Raven Gate — 3 Raven Gate — 4 Raven Gate — 5

Results in numbers

  • 120k

    requests per second at peak

  • <8 ms

    p99 latency

  • 99.98%

    uptime over six months

  • −42%

    infrastructure costs

01

Challenge

The legacy monolithic proxy buckled under peak load, and every routing change required a restart that caused downtime.

02

Solution

We built a Go gateway with distributed rate limiting in Redis, on-the-fly JWT validation and PostgreSQL-backed routes with hot reload, deployed behind NGINX Ingress with Kubernetes autoscaling.

03

Result

The gateway handles peak load without degradation, routes change with zero downtime and Prometheus metrics give a full picture of the traffic.

Key features

  • 01 Distributed rate limiting per key, IP and plan
  • 02 Hot route reload without dropping connections
  • 03 Service-to-service mTLS and key rotation
  • 04 Canary releases with automatic rollback

Architecture

  1. Clients web & mobile apps
  2. NGINX Ingress TLS, load balancing
  3. Raven Gate Go, auth, limits
  4. Redis rate counters
  5. Services 30+ microservices

Tech stack

  • Go
  • Redis
  • PostgreSQL
  • Kubernetes
  • NGINX Ingress
  • Prometheus

Timeline

  1. 1 week

    Audit of the current system

  2. 2 weeks

    Architecture & prototype

  3. 7 weeks

    Development & integration

  4. 2 weeks

    Load testing & launch

Need something similar?

Describe your task in a messenger — we will come back with questions, an architecture and an estimate.

Discuss a similar project All projects