Novus Load Balancer

A custom-built Reverse Proxy, Load Balancer, and LRU Cache implemented in Go.
It distributes client requests across backend servers, detects unhealthy nodes, and optimizes performance with caching. Works similar to servers of NGINX.

Features

• Reverse Proxy → Forwards requests from clients to backend servers.
• Load Balancer → Supports:
  • Round Robin
  • Least Connections
• Health Checks → Periodically pings backends and marks them UP/DOWN.
• Custom LRU Cache → Built using a custom doubly linked list for O(1) operations.
• Configurable via YAML → Easy setup of servers, strategies, and cache size.

Project Structure

│── cmd/
│   ├── main.go          # Entry point of the application
│
│── config/
│   ├── config.go        # Loads configuration from file
│   ├── config.yaml      # Configuration file 
│
│── internal/
│   ├── proxy/
│   │   ├── proxy.go     # Reverse proxy logic
│   ├── loadbalancer/
│   │   ├── lb.go        # Load balancing logic
│   ├── cache/
│   │   ├── lru.go       # LRU cache implementation
│
│── pkg/
│   ├── utils.go         # Helper functions
│
|── Documentation.md     # Project Documentation
│── go.mod               # Go module file
│── go.sum               # Dependencies checksum
│── README.md            # Project Overview

Architecture Diagram

Usage

1. Create the config file at config/config.yaml

If you are using the template file, copy config/config.yaml.example to config/config.yaml and then edit values.

Use this structure (must match the keys exactly):

server:
  port: 8080   # Proxy listening port

proxy:
  targets:
    - "http://localhost:5000"
    - "http://localhost:5001"

loadbalancer:
  strategy: "round_robin"   # Options: round_robin, least_connections

cache:
  max_size: 100   # Number of entries in LRU cache

2. Start backend servers (example using Python HTTP server)

python -m http.server 5000
python -m http.server 5001

3. Run the reverse proxy server

go run cmd/main.go

4. Send request

Example request:

curl http://localhost:8080/

Another example with path and query params:

curl "http://localhost:8080/index.html?source=test"

• First request → forwarded to backend.
• Repeated request → served from cache (logs ✅ Cache hit).
• If backend goes down → logged as ❌ DOWN, skipped in routing.

Cache Efficiency Benchmark

Run benchmark tests for the LRU cache:

go test -bench="." -benchmem ./internal/cache

This includes:

• BenchmarkLRUCachePut for write path performance
• BenchmarkLRUCacheGetHit for cache-hit read performance
• BenchmarkLRUCacheGetMiss for cache-miss read performance

See examples in Documentation.md

Tech Stack

• Go (Golang)
• YAML for Configuration