🌐 Distributed API Gateway

A production-grade API Gateway built to meticulously handle reverse proxying, per-API-key rate limiting, asynchronous traffic analytics, and live monitoring via Server-Sent Events.

---

🌍 Live Environment

• 🚀 Admin Dashboard: https://api-gateway-adesh.vercel.app
• 🔌 Gateway API: https://13-127-252-167.sslip.io
• 💖 Health Status: https://13-127-252-167.sslip.io/health

---

🏗 Technical Architecture

System Design Overview

┌─────────────────────────────────────────────────────────────┐
│                       VERCEL EDGE CLOUD                     │
│                                                             │
│  ┌─────────────────┐ ┌────────────────┐ ┌────────────────┐  │
│  │  Admin UI       │ │  Serverless    │ │  Live Charts   │  │
│  │  (Next.js)      │ │  API Routes    │ │  (Recharts)    │  │
│  └─────────────────┘ └────────────────┘ └────────────────┘  │
└────────────────────────────┬────────────────────────────────┘
                             │
                             │ HTTPS / Live SSE Stream
                             ▼
┌─────────────────────────────────────────────────────────────┐
│                        AWS EC2 INSTANCE                     │
│                                                             │
│  ┌───────────────────────────────────────────────────────┐  │
│  │                 CADDY REVERSE PROXY                   │  │
│  │  - SSL/TLS Termination (Let's Encrypt / sslip.io)     │  │
│  │  - Read Timeout: 300s (Preserves SSE connections)     │  │
│  └─────────────────────────┬─────────────────────────────┘  │
│                            │ HTTP: 3001 (Internal)          │
│                            ▼                                │
│  ┌───────────────────────────────────────────────────────┐  │
│  │                  NODE.JS API GATEWAY                  │  │
│  │                                                       │  │
│  │  ┌──────────────┐   ┌──────────────┐   ┌───────────┐  │  │
│  │  │ Rate Limiter │   │ Route Proxy  │   │ Analytics │  │  │
│  │  │ (Lua Config) │   │ (Express)    │   │ Batch API │  │  │
│  │  └──────┬───────┘   └──────┬───────┘   └─────┬─────┘  │  │
│  └─────────┼──────────────────┼─────────────────┼────────┘  │
│            │                  │                 │           │
│  ┌─────────▼───────┐ ┌────────▼───────┐ ┌───────▼────────┐  │
│  │ REDIS DATATIER  │ │ UPSTREAM APIs  │ │ POSTGRES DB    │  │
│  │ - Cache         │ │ - Microservices│ │ - Auth Keys    │  │
│  │ - Atomic Limits │ │                │ │ - Traffic Logs │  │
│  └─────────────────┘ └────────────────┘ └────────────────┘  │
└─────────────────────────────────────────────────────────────┘

---

🚀 Architectural Flow & Decisions

1. Decoupled Architecture (Vercel + AWS)
   The system is split to play to the strengths of both platforms. Vercel hosts the stateless React frontend globally on its Edge CDN. AWS EC2 handles the stateful, high-throughput, long-living Server-Sent Event (SSE) connections that serverless functions would otherwise prematurely terminate.

2. Zero-Config SSL via sslip.io
   Managing HTTPS manually in Node.js is cumbersome. By routing traffic through Caddy using an sslip.io IP-based domain, the system automatically provisions and renews production-grade Let's Encrypt certificates with zero manual DNS mapping.

3. High-Velocity Rate Limiting
   The Gateway enforces strict API Quotas using Redis array-based Lua scripts. This pushes the rate-limit calculation directly into the Redis memory cache, ensuring atomic evaluations and preventing race conditions under high concurrency.

4. Analytics Batch Processing
   To prevent intense load from choking the PostgreSQL database during traffic spikes, the Gateway logs every request immediately into an in-memory queue. A highly optimized Batch Writer flushes this memory buffer to PostgreSQL asynchronously every 5 seconds.

---

💻 Tech Stack

• Dashboard: Next.js 15, React, TailwindCSS v4, Recharts
• Gateway Core: Node.js, Express, http-proxy-middleware, TypeScript
• State & Data: PostgreSQL (node-postgres), Redis (ioredis)
• DevOps: Docker, GitHub Actions, Caddy

---

🛠 Local Setup

Clone the repository and bootstrap the workspaces:

npm install

Boot the persistence databases locally via Docker:

docker compose up -d postgres redis

Run schema migrations and initialize both frontend & backend simultaneously:

# Terminal 1: Backend
npm run migrate:gateway
npm run dev:gateway

# Terminal 2: Frontend
npm run dev:dashboard

---

🚢 Production CI/CD Pipeline

The project implements an automated zero-downtime deployment pipeline via GitHub Actions.

1. Any push to main triggers a GitHub Action runner.
2. The pipeline surgically builds a strictly minimized Node.js container for the Gateway, dynamically stripping all heavy Dashboard dependencies from the Docker build context to create a lightning-fast image.
3. The image is published to GHCR.
4. The runner connects via SSH into the AWS EC2 Host, pulls the latest images, safely runs PostgreSQL migrations via a temporary container, and gracefully spins up the PostgreSQL, Redis, Gateway, and Caddy instances without dropping existing traffic.