Go to Production: VPS, Docker or Cloud — Choosing the Best Deployment

Why Go is an Ideal Candidate for Production, but Deployment Still Decides Everything

You wrote a backend in Go. It flies: compilation into a single binary, minimal memory consumption, goroutines that aren't afraid of load. But there's a nuance: even the fastest code is useless if it sits as dead weight on a local machine. Sooner or later, the question arises: how do you roll this marvel of engineering out to production?

Deploying Go applications is not rocket science, but it's easy to step on a rake here. In this article, we'll break down three main paths: classic VPS, containerization with Docker, and cloud platforms. You'll learn which approach to choose for your task and get ready-made scripts to get started.

Deploying Go on VPS: A Classic That Never Gets Old

A Virtual Private Server (VPS) is your personal root. Full control, no platform restrictions, but all responsibility lies with you. For a Go application, this is often the optimal option: the binary weighs 10–20 MB, starts in milliseconds, and doesn't require complex orchestration.

What You Need for Deploying Go on VPS

\n
• Server with Linux (Ubuntu 22.04 / Debian 11 — the classics)
\n
• SSH access
\n
• Go (optional, if compiling on the server)
\n
• Systemd for process management
\n
• Nginx or Caddy as a reverse proxy
\n

Example: Deploying a Go Server via Systemd

Let's say you have a simple application main.go that listens on port 8080. Compile it on your machine for the target architecture:

GOOS=linux GOARCH=amd64 go build -o app main.go

Copy the binary to the server:

scp app user@your-server:/opt/app/

Create a systemd unit at /etc/systemd/system/app.service:

[Unit]\nDescription=Go App\nAfter=network.target\n\n[Service]\nExecStart=/opt/app/app\nWorkingDirectory=/opt/app\nRestart=always\nUser=www-data\nGroup=www-data\nEnvironment=PORT=8080\n\n[Install]\nWantedBy=multi-user.target

Start and enable auto-start:

sudo systemctl daemon-reload\nsudo systemctl enable --now app

Now your Go application runs as a daemon. Add Nginx for proxying from port 80:

server {\n    listen 80;\n    server_name yourdomain.com;\n\n    location / {\n        proxy_pass http://127.0.0.1:8080;\n        proxy_set_header Host $host;\n        proxy_set_header X-Real-IP $remote_addr;\n    }\n}

VPS Pros: full control, low cost (from $5/month), transparency. Cons: manual updates, no autoscaling, need to configure security yourself.

Docker for Go: Containerization That Simplifies Life

If you work in a team or plan microservices, Docker is your choice. Go and Docker are a perfect pair: multi-stage builds produce images sized 5–15 MB, and static compilation allows you to do without a base image.

Dockerfile for a Go Application

Here is a minimal but production-ready Dockerfile:

# Stage 1: build\nFROM golang:1.22-alpine AS builder\nWORKDIR /app\nCOPY go.mod go.sum ./\nRUN go mod download\nCOPY . .\nRUN CGO_ENABLED=0 GOOS=linux go build -o app .\n\n# Stage 2: minimal image\nFROM alpine:latest\nRUN apk --no-cache add ca-certificates\nWORKDIR /root/\nCOPY --from=builder /app/app .\nEXPOSE 8080\nCMD ["./app"]

Build and run:

docker build -t go-app .\ndocker run -d -p 8080:8080 --name go-app go-app

Docker Compose for a Full Setup

When the application needs a database or Redis, use docker-compose.yml:

version: '3.8'\nservices:\n  app:\n    build: .\n    ports:\n      - "8080:8080"\n    environment:\n      - DB_HOST=db\n      - DB_PORT=5432\n    depends_on:\n      - db\n  db:\n    image: postgres