A production-style HTTP load balancer in Go with strategy switching, health checks, metrics, and integrated request rate limiting.

Current stable version: v1.0.0

License: MIT (redistributable)

Features

• Multiple balancing strategies:
  • round-robin
  • least-connections
  • weighted (smooth weighted round robin)
• Active health checks that mark unhealthy backends out of rotation
• YAML/JSON config loading
• Request logging
• Basic Prometheus-style metrics endpoint at /metrics
• Optional integration with github.com/codestorm1875/ratelimiter as middleware

Architecture

flowchart LR
  C[Client] --> RL[Rate Limiter Middleware]
  RL --> LB[Load Balancer Handler]
  LB --> S{Strategy Engine}
  S --> A[Backend A]
  S --> B[Backend B]
  S --> N[Backend N]
  HC[Health Checker] --> A
  HC --> B
  HC --> N
  LB --> M[Metrics Registry]
  M --> P["/metrics"]

Loading

Request path summary:

1. Incoming traffic enters optional rate limiting middleware.
2. The balancer selects from healthy backends using configured strategy.
3. Active health checks continuously mark backends in or out of rotation.
4. Request and backend counters are exposed at /metrics.

Run

make run

For local development, start two mock backends in separate terminals first:

make backend-a
make backend-b

Then start the load balancer:

make run

Note: config.example.yaml has rate limiting disabled by default for smoother local smoke tests. Enable rate_limit.enabled: true when you want to validate throttling behavior.

Config

Copy config.example.yaml and adjust backend URLs.

The same config fields work for JSON and YAML files.

Endpoints

• / proxied to selected upstream backend
• /metrics exposes counters and per-backend state
• /health local balancer health endpoint

Testing

Standard workflow:

make check

Useful focused commands:

make test
make test-integration
make bench

Direct Go commands remain available:

go mod tidy
go test ./...
go test -run TestIntegrationFailoverAndRecovery ./internal/lb

Metrics Provided

• lb_requests_total
• lb_errors_total
• lb_ratelimited_total
• lb_active_requests
• lb_backend_up{name="..."}
• lb_backend_active_connections{name="..."}
• lb_backend_requests_total{name="..."}
• lb_backend_errors_total{name="..."}

Design Decisions and Tradeoffs

This project demonstrates a layered architecture:

1. Traffic shaping at the edge with strategy-based load distribution
2. Automatic backend lifecycle management via active health checks
3. Abuse control with fairness-aware rate limiting middleware
4. Observability through structured logs and scrape-friendly metrics

Versioning And Releases

• Semantic versioning is used for stable releases.
• The current release tag is v1.0.0.
• Release notes are tracked in CHANGELOG.md.