Skip to content

Open Codelabs GCP (Google Cloud Platform) Deployment Guide

This document describes several ways to deploy Open Codelabs on GCP.

GCP deployment options

Depending on your architecture, there are three primary approaches.

Firebase is a developer-friendly suite of GCP services. The project already supports Firebase mode, so you can deploy without managing servers.

  • Hosting: frontend (SvelteKit)
  • Firestore: codelab data and user info
  • Realtime Database: chat and progress sharing
  • Storage: image uploads
  • Auth: Google login

See the detailed guide: DEPLOY_FIREBASE.md.

2. Cloud Run (container-based)

Build the backend (Rust) and frontend as containers and deploy them separately. This is suitable when you want more control over the infrastructure.

Architecture

  • Frontend: SvelteKit (Node.js/Bun) -> Cloud Run service 1
  • Backend: Rust Axum -> Cloud Run service 2
  • Database:
  • Simple: SQLite on a Cloud Run local volume. Note: Cloud Run is stateless, so the SQLite file resets on restart. For non-temporary usage, use the recommended option below.
  • Recommended: switch to Cloud SQL (PostgreSQL/MySQL) or use Firebase mode.

Deployment steps

1. Configure the GCP project and login 

gcloud auth login
gcloud config set project [YOUR_PROJECT_ID]

2. Create Artifact Registry 

gcloud artifacts repositories create open-codelabs \
    --repository-format=docker --location=asia-northeast3

3. Build and push images 

# Backend build
cd backend
gcloud builds submit --tag asia-northeast3-docker.pkg.dev/[PROJECT_ID]/open-codelabs/backend:latest .

# Frontend build
cd ../frontend
gcloud builds submit --tag asia-northeast3-docker.pkg.dev/[PROJECT_ID]/open-codelabs/frontend:latest .

4. Deploy Cloud Run services - Backend: 

gcloud run deploy backend \
    --image asia-northeast3-docker.pkg.dev/[PROJECT_ID]/open-codelabs/backend:latest \
    --platform managed --region asia-northeast3 --allow-unauthenticated \
    --set-env-vars "DATABASE_URL=sqlite:data/sqlite.db?mode=rwc,ADMIN_ID=admin,ADMIN_PW=admin123"
- Frontend: 
gcloud run deploy frontend \
    --image asia-northeast3-docker.pkg.dev/[PROJECT_ID]/open-codelabs/frontend:latest \
    --platform managed --region asia-northeast3 --allow-unauthenticated \
    --set-env-vars "VITE_API_URL=[BACKEND_SERVICE_URL]"

3. Compute Engine (traditional VM)

Deploy with Docker Compose on a GCP VM.

  1. Create a GCE instance (Debian or Ubuntu)
  2. Install Docker and Docker Compose
  3. Clone the repo and run docker-compose.yml 
    docker compose up -d --build

Which option should you choose?

Feature Firebase (Serverless) Cloud Run (Containers) Compute Engine (VM)
Difficulty Very low Medium Medium
Cost Usage based (generous free tier) Usage based Fixed (instance time)
Realtime Firebase RTDB (very stable) WebSocket (sticky session required) WebSocket (full control)
Maintenance Almost none Low High

Conclusion: For a first deployment or small workshops, Firebase is strongly recommended. If you need custom server behavior or want to go deeper with GCP, consider Cloud Run.