## TLDR Migrating from Heroku to DigitalOcean can reduce infrastructure costs by 60-80% with App Platform, or 90%+ with Coolify self-hosted. DigitalOcean's App Platform provides a comparable developer experience to Heroku with git-based deployments, auto-scaling, and zero-downtime deploys. For maximum savings, Coolify offers a self-hosted alternative running on a single $24/month Droplet supporting multiple apps. Combined with Managed Databases, Spaces (S3-compatible storage), and managed services, you get Heroku-like simplicity at a fraction of the cost. This guide walks through both paths with production migration strategies and minimal downtime. --- ## Why Migrate from Heroku to DigitalOcean? ### Cost Comparison: Real Numbers **Typical Production Heroku Setup**: - 2× Performance-M dynos (web): $50/month each = $100/month - 1× Performance-M dyno (worker): $50/month - Standard-0 Postgres: $50/month - Premium-0 Redis: $15/month - Review apps (2 active): $30/month - **Total: $245/month** For a more demanding workload: - 4× Performance-L dynos (web): $500/month each = $2,000/month - 2× Performance-L dynos (workers): $1,000/month - Standard-4 Postgres: $200/month - Premium-5 Redis: $350/month - **Total: $3,550/month** **Equivalent DigitalOcean Setup (Basic)**: - App Platform: 2× Professional instances ($24/month each) = $48/month - Managed Database (PostgreSQL): Basic plan = $15/month - Managed Redis: Basic plan = $15/month - Spaces (object storage): $5/month + transfer - **Total: $83/month** - **Savings: 66% ($162/month)** **Equivalent DigitalOcean Setup (High-Performance)**: - App Platform: 4× Professional instances ($48/month for larger) = $192/month - App Platform workers: 2× instances = $96/month - Managed Database: Production plan (4GB RAM) = $60/month - Managed Redis: Production plan = $60/month - Spaces: $5/month - **Total: $413/month** - **Savings: 88% ($3,137/month)** Beyond cost, you gain: - More predictable pricing (no dyno sleep, clearer resource limits) - Better performance per dollar (dedicated resources, not shared containers) - Infrastructure flexibility (VPCs, Kubernetes, Droplets when needed) - S3-compatible object storage included **Alternative: Coolify on DigitalOcean Droplet (Even Cheaper)**: [Coolify](https://coolify.io) is an open-source, self-hostable Heroku/Netlify alternative that you can deploy on a single DigitalOcean Droplet. It provides git-based deployments, automatic SSL, and built-in database management. - 1× Droplet (4GB RAM, 2 vCPUs): $24/month - PostgreSQL (on same Droplet): $0 (self-hosted) - Redis (on same Droplet): $0 (self-hosted) - Object storage: Spaces $5/month (optional, can use Droplet storage) - **Total: $24-29/month** - **Savings: 88-90% vs Heroku** ($216-221/month saved) **Coolify Trade-offs**: - ✅ **Pros**: Lowest cost, full control, Docker-based deployments, multiple apps per server - ⚠️ **Cons**: Self-managed (you handle backups, updates, scaling), single point of failure (unless you set up HA) - 🎯 **Best for**: Small teams (<5 apps), budget-conscious startups, developers comfortable with server management **When to choose Coolify over App Platform**: - You're running 3+ small apps (share one Droplet) - You want maximum cost savings and don't mind managing servers - Your apps fit comfortably on a single server (no need for auto-scaling yet) - You're comfortable with Docker and Linux administration --- ## Migration Strategy: Zero-Downtime Approach ### Phase 1: Parallel Infrastructure (Week 1) Run DigitalOcean infrastructure alongside Heroku without switching traffic. **Goal**: Validate that DigitalOcean setup works with production data. **Steps**: 1. Set up DigitalOcean Managed Database (PostgreSQL/MySQL) 2. Configure replication from Heroku Postgres to DigitalOcean 3. Deploy application to App Platform (no public traffic yet) 4. Test functionality with read-replica data 5. Monitor performance and identify issues **Risk**: Low. Heroku remains primary, DigitalOcean is shadow environment. ### Phase 2: Database Migration (Week 2) Move database to DigitalOcean with minimal downtime. **Strategy**: Use logical replication + cutover window. **Steps**: 1. Set up continuous replication (Heroku → DigitalOcean) 2. Let replication catch up (monitor lag) 3. Schedule maintenance window (typically 5-15 minutes) 4. Stop writes to Heroku 5. Wait for replication to fully sync 6. Update Heroku app DATABASE_URL to point to DigitalOcean 7. Resume traffic **Downtime**: 5-15 minutes (writes only, reads can continue) ### Phase 3: Application Migration (Week 3) Move application traffic to App Platform. **Strategy**: Gradual traffic shift using DNS. **Steps**: 1. Deploy app to App Platform with DigitalOcean database 2. Set DNS TTL to 60 seconds 3. Add App Platform URL as secondary A record (10% traffic) 4. Monitor errors, latency, throughput 5. Gradually increase traffic: 25% → 50% → 100% 6. Decommission Heroku dynos **Rollback**: Simple DNS change back to Heroku. ### Phase 4: Supporting Services (Week 4) Migrate Redis, object storage, background jobs. **Steps**: 1. Set up Managed Redis on DigitalOcean 2. Migrate Spaces (or keep existing S3, update credentials) 3. Update worker processes to use DigitalOcean Redis 4. Move scheduled jobs to App Platform workers **Total migration time**: 3-4 weeks with minimal risk. --- ## Technical Implementation: Component by Component ### 1. DigitalOcean App Platform **What it is**: Platform-as-a-Service similar to Heroku. Git-based deployments, auto-scaling, managed runtime. **How it compares to Heroku**: - **Buildpacks**: Supports Docker, Node.js, Python, Ruby, Go, PHP out of box - **Deployments**: Git push or GitHub integration (like Heroku) - **Scaling**: Horizontal auto-scaling based on CPU/memory - **Zero-downtime**: Rolling deployments (like Heroku) - **Review apps**: Preview environments from PRs **Key differences**: - More explicit resource limits (CPU, RAM clearly defined) - Lower base cost ($5/month starter vs $7/month Heroku Eco) - No dyno sleeping (all instances stay running) - Better observability (built-in metrics, no add-on needed) #### Setup Example: Node.js API **Option 1: GitHub Integration (Recommended)** 1. Create `.do/app.yaml` in your repository: ```yaml name: my-api region: nyc services: - name: web github: repo: yourorg/your-repo branch: main deploy_on_push: true build_command: npm install && npm run build run_command: npm start instance_count: 2 instance_size_slug: professional-xs # $12/month per instance http_port: 3000 health_check: http_path: /health initial_delay_seconds: 10 period_seconds: 10 envs: - key: NODE_ENV value: production - key: DATABASE_URL type: SECRET - key: REDIS_URL type: SECRET - name: worker github: repo: yourorg/your-repo branch: main build_command: npm install run_command: npm run worker instance_count: 1 instance_size_slug: basic-xs # $5/month envs: - key: NODE_ENV value: production - key: DATABASE_URL type: SECRET - key: REDIS_URL type: SECRET databases: - name: production-db engine: PG version: "15" production: true cluster_name: my-db-cluster ``` 2. Deploy via CLI: ```bash # Install doctl (DigitalOcean CLI) brew install doctl # Authenticate doctl auth init # Create app from spec doctl apps create --spec .do/app.yaml # Or deploy via UI: Apps → Create → Import from GitHub ``` **Option 2: Dockerfile Deployment** If you have custom Docker setup: ```yaml services: - name: web dockerfile_path: Dockerfile source_dir: / instance_count: 2 instance_size_slug: professional-xs http_port: 8080 ``` **Auto-Scaling Configuration**: ```yaml services: - name: web instance_count: 2 autoscaling: min_instance_count: 2 max_instance_count: 10 metrics: cpu: percent: 75 # Scale up when CPU > 75% ``` #### Environment Variables and Secrets **Set via CLI**: ```bash # Get app ID doctl apps list # Set environment variable doctl apps update $APP_ID --set-env="KEY=value" # Set secret (encrypted) doctl apps update $APP_ID --set-env="DATABASE_URL=postgresql://..." --encrypt ``` **Set via UI**: Apps → Settings → Environment Variables **Best practice**: Use App Platform's managed database integration for automatic DATABASE_URL injection. --- ### Alternative: Coolify Self-Hosted Setup If you're willing to manage your own server in exchange for dramatically lower costs, Coolify offers a compelling path. This open-source platform runs entirely on a single DigitalOcean Droplet and can host multiple applications for as little as $24/month — a fraction of what you'd pay on Heroku or even App Platform. #### What is Coolify? Think of Coolify as a self-hosted Heroku. It gives you the same git-push deployment experience you're used to, but everything runs on infrastructure you control. Behind the scenes, it uses Docker for containerization, Traefik for routing and SSL termination, and provides a clean web UI for managing everything. The platform supports GitHub, GitLab, and Bitbucket repositories with automatic deployments on push. SSL certificates are handled automatically through Let's Encrypt, and you get built-in support for PostgreSQL, MySQL, MongoDB, and Redis databases. Whether you're deploying static sites, APIs, or full-stack applications, Coolify handles the orchestration while you maintain full control over the underlying infrastructure. #### Setting Up Coolify The setup process takes about 15 minutes from start to finish. You'll start by spinning up a fresh Ubuntu Droplet, run Coolify's installer, configure DNS, and deploy your first app. Here's the step-by-step walkthrough. **Creating Your Droplet** Start with a clean Ubuntu 22.04 server. For 1-3 small applications, a 4GB RAM Droplet ($24/month) is plenty. If you're running 3-5 medium-traffic apps or processing background jobs, step up to 8GB ($48/month). High-traffic setups with 5-10 apps work well on 16GB ($96/month). ```bash # Create a Droplet via CLI doctl compute droplet create coolify-server \\ --image ubuntu-22-04-x64 \\ --size s-2vcpu-4gb \\ --region nyc3 \\ --ssh-keys YOUR_SSH_KEY_ID # Or use the DigitalOcean UI: # - Ubuntu 22.04 LTS # - 4GB RAM / 2 vCPUs (starts at $24/month) # - NYC3 or any region ``` [Sign up for DigitalOcean](https://www.jdoqocy.com/click-101674709-15836238) and receive $200 in credits to test Coolify risk-free. **Installing Coolify** Once your Droplet is running, SSH in and run Coolify's installer. The script handles all dependencies — Docker, Docker Compose, and the Coolify control plane. Installation takes 5-10 minutes depending on your connection speed. ```bash # SSH into Droplet ssh root@your-droplet-ip # Install Coolify (takes 5-10 minutes) curl -fsSL https://cdn.coollabs.io/coolify/install.sh | bash # After installation, access Coolify at: # http://your-droplet-ip:8000 ``` After installation completes, you'll access Coolify's web interface at `http://your-droplet-ip:8000`. The first time you log in, you'll create an admin account and set up your email for SSL certificate notifications. **Configuring DNS** Before deploying apps, point your domain to the Droplet's IP address: ``` # A record coolify.yourdomain.com → your-droplet-ip # Wildcard for subdomains (optional) *.coolify.yourdomain.com → your-droplet-ip ``` The wildcard record is optional but recommended — it lets you deploy multiple apps on different subdomains without manually creating DNS records each time. **Deploying Your First Application** The deployment flow in Coolify feels familiar if you've used Heroku. Start by creating a project (think of it as a workspace), then add a new application. Connect your GitHub, GitLab, or Bitbucket repository, and Coolify will analyze your code to detect the framework. For most frameworks, Coolify uses Nixpacks (similar to Heroku's buildpacks) to automatically detect and build your app. If you have a Dockerfile, it'll use that instead. Set your environment variables, specify your custom domain, and hit deploy. Coolify pulls your code, builds it, starts the container, and provisions an SSL certificate — all automatically. **Example deployment (Node.js)**: ```yaml # Coolify auto-detects from package.json, but you can customize: # Environment Variables (in Coolify UI): NODE_ENV=production PORT=3000 DATABASE_URL=postgresql://user:pass@localhost:5432/myapp REDIS_URL=redis://localhost:6379 # Build Command (optional override): npm run build # Start Command: npm start ``` #### Adding Databases with Coolify One of Coolify's best features is built-in database management. Instead of provisioning separate managed databases, you can deploy PostgreSQL, MySQL, MongoDB, or Redis directly on your Droplet through the same web interface. **Setting up PostgreSQL** takes about 30 seconds. Navigate to Resources → New Database → PostgreSQL, set your database name and credentials, and Coolify spins up a containerized PostgreSQL instance. The connection string is generated automatically, so you can copy it directly into your application's environment variables. **Redis works the same way**. Create a new Redis resource, choose version 7.x (recommended for stability), and Coolify handles persistence configuration and automatic restarts. Your apps connect via `redis://localhost:6379`. **Backing up your databases** is critical since you're managing the infrastructure. Here's a simple backup script that dumps PostgreSQL and ships it to DigitalOcean Spaces: ```bash # Coolify stores data in Docker volumes # Backup PostgreSQL: docker exec coolify-postgres pg_dump -U postgres myapp > backup.sql # Backup Redis: docker exec coolify-redis redis-cli SAVE docker cp coolify-redis:/data/dump.rdb ./redis-backup.rdb # Automate with cron: 0 2 * * * /root/backup-databases.sh ``` #### Migrating from Heroku to Coolify The migration process from Heroku follows the same pattern as migrating to App Platform, but with a few extra manual steps since you're managing the infrastructure. **Export your Heroku data first**. Capture a fresh database backup and download your environment variables: ```bash # Export database heroku pg:backups:capture -a myapp heroku pg:backups:download -a myapp # Get environment variables heroku config -a myapp --shell > .env.production ``` **Restore your database on Coolify**. SSH into your Droplet and restore the dump directly into the Coolify-managed PostgreSQL container: ```bash # SSH into Coolify Droplet ssh root@coolify-server # Restore PostgreSQL dump docker exec -i coolify-postgres psql -U postgres myapp < latest.dump # Verify data docker exec coolify-postgres psql -U postgres myapp -c "SELECT count(*) FROM users;" ``` **Deploy your application through the Coolify UI**. Create the app, connect your repository, and paste in your environment variables from the `.env.production` file you exported. Deploy to a temporary subdomain first to test everything works before switching DNS. Once you've verified the app works correctly, update your DNS records to point to the Coolify Droplet. Coolify will automatically request and install an SSL certificate from Let's Encrypt. This typically takes 1-2 minutes. **Decommission Heroku only after monitoring for 24-48 hours**. Put Heroku in maintenance mode while you verify everything in production: ```bash heroku maintenance:on -a myapp # Monitor Coolify for 24-48 hours # Then delete Heroku app ``` If you spot any issues during this monitoring window, you can quickly revert by turning off maintenance mode on Heroku. #### Cost Comparison: Coolify vs App Platform vs Heroku **Scenario: 3 small apps + PostgreSQL + Redis** | **Platform** | **Monthly Cost** | **Notes** | |--------------|------------------|------------| | **Heroku** | $735/month | 3× Performance-M ($150 each) + 3× Standard-0 Postgres ($150) + 3× Premium-0 Redis ($45) | | **App Platform** | $249/month | 3× Professional-XS apps ($72) + 3× Managed DB Basic ($45) + 3× Managed Redis ($45) + Spaces ($5) | | **Coolify** | $24-48/month | 1× Droplet 4-8GB ($24-48) + self-hosted databases (no extra cost) | **Savings**: Coolify is **94-97% cheaper** than Heroku for multi-app setups. #### Coolify Best Practices Once you're running on Coolify, a few operational practices will save you headaches down the road. **Set up automated backups immediately**. Here's a script that runs nightly and ships database dumps to DigitalOcean Spaces: ```bash # /root/backup-to-spaces.sh #!/bin/bash DATE=$(date +%Y%m%d) docker exec coolify-postgres pg_dumpall -U postgres | gzip > /tmp/db-$DATE.sql.gz s3cmd put /tmp/db-$DATE.sql.gz s3://my-backups/coolify/ ``` Add this to cron with `0 2 * * * /root/backup-to-spaces.sh` to run at 2 AM daily. Store at least 7 days of backups and test your restore process quarterly. **Monitor your applications** using Uptime Kuma, which you can also deploy through Coolify. It's lightweight, provides downtime alerts, and integrates with Slack, Discord, or email notifications. **Lock down your server** with UFW firewall. Only expose SSH (port 22), HTTP (80), and HTTPS (443). Keep Coolify updated with `coolify update` every month or when security patches are released. Use strong, randomly generated passwords for all database credentials. **Plan your exit strategy**. When you outgrow a single Droplet — typically around 10K-50K requests/minute or when you need multi-region deployment — you can migrate to App Platform or Kubernetes. The containerized nature of Coolify makes this transition straightforward. #### When Coolify is the Right Choice Coolify shines in specific scenarios. It's ideal when you're running multiple applications and want to consolidate them on shared infrastructure — the cost savings compound quickly. You'll need basic Linux comfort (SSH, Docker concepts, reading logs), but you don't need to be a sysadmin. If you're currently spending $500+/month on Heroku across several apps, Coolify can cut that to $24-$96/month. The traffic sweet spot is 1K-10K requests/minute on an 8GB Droplet. Beyond that, you'll want to either scale vertically to 16GB+ or consider moving to App Platform for horizontal scaling. **Avoid Coolify if** you need enterprise SLAs, 24/7 vendor support, or multi-region redundancy out of the box. It's also not the right choice if you've never SSH'd into a server before — there's a learning curve. And while you can vertically scale Droplets quickly, instant horizontal auto-scaling isn't available like it is with App Platform. #### Coolify + App Platform Hybrid Many teams find the sweet spot by using both platforms. Run your staging and development environments on Coolify for $24/month, and keep production on App Platform with managed services for $83/month. This gives you cost-effective testing environments and production reliability. Total cost: $107/month compared to $490/month on Heroku for equivalent staging + production environments. That's 78% savings while maintaining the safety of managed infrastructure where it matters most. --- ### 2. Managed Databases DigitalOcean's Managed Databases offer PostgreSQL, MySQL, MongoDB, and Redis with automated backups, point-in-time recovery, read replicas, and connection pooling. #### PostgreSQL Setup **Create via CLI**: ```bash # Create production database cluster doctl databases create production-postgres \ --engine pg \ --version 15 \ --region nyc3 \ --size db-s-2vcpu-4gb \ --num-nodes 1 # Get connection details doctl databases connection production-postgres # Create database and user doctl databases db create production-postgres myapp doctl databases user create production-postgres myapp-user ``` **Connection Pooling** (recommended for production): ```bash # Create connection pool doctl databases pool create production-postgres myapp-pool \ --db myapp \ --user myapp-user \ --size 25 \ --mode transaction ``` **Connection string format**: ``` # Direct connection postgresql://username:password@host:25060/database?sslmode=require # Pooled connection (recommended) postgresql://username:password@host:25061/database?sslmode=require ``` #### Migrating Data from Heroku Postgres **Option 1: Logical Replication (Zero Downtime)** Best for databases >10GB with minimal downtime requirements. ```bash # 1. On Heroku Postgres, enable logical replication heroku pg:psql -a myapp ALTER SYSTEM SET wal_level = logical; SELECT pg_reload_conf(); # 2. Create publication on source CREATE PUBLICATION heroku_pub FOR ALL TABLES; # 3. On DigitalOcean database, create subscription CREATE SUBSCRIPTION do_sub CONNECTION 'postgresql://heroku_host:5432/database' PUBLICATION heroku_pub; # 4. Monitor replication lag SELECT * FROM pg_stat_subscription; # 5. When lag is near zero, stop writes and switch ``` **Option 2: pg_dump/pg_restore (Simpler, Downtime Required)** Best for databases <10GB or when downtime is acceptable. ```bash # 1. Put Heroku app in maintenance mode heroku maintenance:on -a myapp # 2. Create dump from Heroku heroku pg:backups:capture -a myapp heroku pg:backups:download -a myapp # 3. Restore to DigitalOcean pg_restore --verbose --clean --no-acl --no-owner \ -h do-host -U do-user -d myapp latest.dump # 4. Verify data integrity psql -h do-host -U do-user -d myapp -c "SELECT count(*) FROM users;" # 5. Update DATABASE_URL in App Platform # 6. Deploy and test # 7. Turn off Heroku maintenance mode ``` #### Backup Configuration DigitalOcean automatically backs up databases daily: ```bash # List available backups doctl databases backups list production-postgres # Restore from backup doctl databases backups restore production-postgres backup-id # Fork database to new cluster (for testing) doctl databases fork production-postgres test-postgres \ --restore-from-timestamp "2026-03-01T10:30:00Z" ``` **Point-in-Time Recovery**: Available on clusters $40/month and above. Allows restore to any point within 7-day window. #### Cost Comparison | **Heroku Postgres** | **DigitalOcean Managed DB** | **Savings** | |---------------------|-----------------------------|--------------| | Hobby-dev: Free (10K rows limit) | Basic (1GB RAM, 10GB disk): $15/month | N/A | | Mini: $5/month (10M rows) | Same as above | N/A | | Standard-0: $50/month (64GB storage) | Professional (2GB RAM, 25GB): $30/month | 40% | | Standard-4: $200/month (256GB storage) | Professional (4GB RAM, 80GB): $60/month | 70% | | Premium-5: $350/month (512GB storage) | Professional (8GB RAM, 160GB): $120/month | 66% | --- ### 3. Managed Redis DigitalOcean's Managed Redis offers high-performance caching and session storage with automated failover. #### Setup ```bash # Create Redis cluster doctl databases create production-redis \ --engine redis \ --version 7 \ --region nyc3 \ --size db-s-1vcpu-1gb \ --num-nodes 1 # Get connection details doctl databases connection production-redis ``` **Connection string format**: ``` redis://username:password@host:25061?ssl=true ``` #### Migration from Heroku Redis **Option 1: Application-Level Migration** (Recommended) Let cache warm up naturally after switching: ```bash # 1. Deploy app with REDIS_URL pointing to DigitalOcean # 2. Cache will repopulate on cache misses # 3. No data migration needed for true caching ``` **Option 2: redis-cli DUMP/RESTORE** (For persistent data): ```bash # Install redis-cli brew install redis # macOS # Export from Heroku heroku redis:cli -a myapp SAVE # Force snapshot BGSAVE # Background save # Use redis-copy tool for migration npm install -g redis-copy redis-copy \ --src redis://heroku-redis-url \ --dst rediss://do-redis-url ``` #### Eviction Policies Configure via DigitalOcean UI or CLI: ```bash # Set maxmemory policy doctl databases options set production-redis \ --config maxmemory-policy=allkeys-lru ``` Common policies: - `allkeys-lru`: Evict least recently used keys (recommended for caching) - `volatile-lru`: Evict LRU keys with TTL set - `noeviction`: Return errors when memory full (for queues) #### Cost Comparison | **Heroku Redis** | **DigitalOcean Redis** | **Savings** | |------------------|------------------------|--------------| | Mini: $15/month (25MB) | Basic (1GB RAM): $15/month | 0% but 40× capacity | | Premium-0: $15/month (100MB) | Basic (1GB RAM): $15/month | 0% but 10× capacity | | Premium-5: $350/month (4GB) | Professional (4GB RAM): $60/month | 83% | --- ### 4. Spaces (Object Storage) Spaces is DigitalOcean's S3-compatible object storage. Fully compatible with AWS SDK, making migration from S3 trivial. #### Creating a Space ```bash # Create Space doctl compute spaces create myapp-production \ --region nyc3 # Generate API keys doctl compute spaces keys create myapp-spaces-key ``` #### S3 SDK Configuration **Node.js (AWS SDK v3)**: ```javascript import { S3Client, PutObjectCommand } from '@aws-sdk/client-s3'; const s3Client = new S3Client({ endpoint: 'https://nyc3.digitaloceanspaces.com', region: 'us-east-1', // Required but ignored credentials: { accessKeyId: process.env.SPACES_KEY, secretAccessKey: process.env.SPACES_SECRET, }, }); // Upload file await s3Client.send(new PutObjectCommand({ Bucket: 'myapp-production', Key: 'uploads/avatar.jpg', Body: fileBuffer, ACL: 'public-read', // Or 'private' })); // Public URL format const publicUrl = `https://myapp-production.nyc3.digitaloceanspaces.com/uploads/avatar.jpg`; // CDN URL (if enabled) const cdnUrl = `https://myapp-production.nyc3.cdn.digitaloceanspaces.com/uploads/avatar.jpg`; ``` **Ruby (aws-sdk-s3)**: ```ruby require 'aws-sdk-s3' s3 = Aws::S3::Resource.new( endpoint: 'https://nyc3.digitaloceanspaces.com', access_key_id: ENV['SPACES_KEY'], secret_access_key: ENV['SPACES_SECRET'], region: 'us-east-1' ) obj = s3.bucket('myapp-production').object('uploads/avatar.jpg') obj.upload_file('/path/to/file.jpg', acl: 'public-read') puts obj.public_url ``` #### Migrating from AWS S3 **Option 1: aws-cli sync**: ```bash # Install aws-cli if not present brew install awscli # Sync from S3 to Spaces AWS_ACCESS_KEY_ID=$SPACES_KEY \ AWS_SECRET_ACCESS_KEY=$SPACES_SECRET \ aws s3 sync s3://my-heroku-bucket/ s3://myapp-production/ \ --endpoint-url https://nyc3.digitaloceanspaces.com ``` **Option 2: rclone (for large datasets)**: ```bash # Install rclone brew install rclone # Configure source (S3) rclone config create s3-source s3 \ access_key_id=$AWS_ACCESS_KEY \ secret_access_key=$AWS_SECRET_KEY # Configure destination (Spaces) rclone config create do-spaces s3 \ access_key_id=$SPACES_KEY \ secret_access_key=$SPACES_SECRET \ endpoint=nyc3.digitaloceanspaces.com # Sync with progress rclone sync s3-source:my-bucket do-spaces:myapp-production --progress ``` #### CDN Integration Enable built-in CDN (free) for faster global delivery: ```bash # Enable CDN via UI: Spaces → Settings → CDN # CDN endpoint: https://myapp-production.nyc3.cdn.digitaloceanspaces.com ``` Benefits: - Free CDN (included in Spaces pricing) - Automatic TLS/SSL - Global edge caching - No Cloudflare setup needed #### Cost Comparison | **Service** | **Pricing** | |-------------|-------------| | **DigitalOcean Spaces** | $5/month for 250GB + 1TB transfer
$0.02/GB over 250GB storage
$0.01/GB over 1TB transfer | | **AWS S3 (us-east-1)** | $0.023/GB storage
$0.09/GB transfer
Minimum ~$10-20/month for typical app | | **Heroku + S3** | Must use external S3 + egress fees | **Example**: 100GB storage + 500GB transfer/month: - **Spaces**: $5/month (included in base) - **AWS S3**: $2.30 (storage) + $45 (transfer) = $47.30/month - **Savings**: 89% --- ## Infrastructure as Code Managing DigitalOcean resources via Terraform ensures reproducibility and version control. ### Terraform Example: Full Stack **main.tf**: ```hcl terraform { required_providers { digitalocean = { source = "digitalocean/digitalocean" version = "~> 2.34" } } } provider "digitalocean" { token = var.do_token } # VPC for private networking resource "digitalocean_vpc" "main" { name = "production-vpc" region = "nyc3" ip_range = "10.10.0.0/16" } # PostgreSQL Database resource "digitalocean_database_cluster" "postgres" { name = "production-postgres" engine = "pg" version = "15" size = "db-s-2vcpu-4gb" region = "nyc3" node_count = 1 private_network_uuid = digitalocean_vpc.main.id } resource "digitalocean_database_db" "app" { cluster_id = digitalocean_database_cluster.postgres.id name = "myapp" } resource "digitalocean_database_user" "app" { cluster_id = digitalocean_database_cluster.postgres.id name = "myapp-user" } # Redis Cache resource "digitalocean_database_cluster" "redis" { name = "production-redis" engine = "redis" version = "7" size = "db-s-1vcpu-1gb" region = "nyc3" node_count = 1 private_network_uuid = digitalocean_vpc.main.id } # Spaces Bucket resource "digitalocean_spaces_bucket" "uploads" { name = "myapp-production" region = "nyc3" cors_rule { allowed_headers = ["*"] allowed_methods = ["GET", "PUT", "POST"] allowed_origins = ["https://myapp.com"] max_age_seconds = 3000 } } # App Platform App resource "digitalocean_app" "web" { spec { name = "myapp" region = "nyc" service { name = "web" instance_count = 2 instance_size_slug = "professional-xs" github { repo = "myorg/myapp" branch = "main" deploy_on_push = true } env { key = "DATABASE_URL" value = digitalocean_database_cluster.postgres.uri type = "SECRET" } env { key = "REDIS_URL" value = digitalocean_database_cluster.redis.uri type = "SECRET" } env { key = "SPACES_KEY" value = var.spaces_key type = "SECRET" } } } } output "app_live_url" { value = digitalocean_app.web.live_url } output "database_uri" { value = digitalocean_database_cluster.postgres.uri sensitive = true } ``` **Apply infrastructure**: ```bash # Initialize terraform init # Plan changes terraform plan -var="do_token=$DIGITALOCEAN_TOKEN" # Apply terraform apply -var="do_token=$DIGITALOCEAN_TOKEN" ``` --- ## CI/CD with GitHub Actions Automate deployments to DigitalOcean App Platform. **.github/workflows/deploy.yml**: ```yaml name: Deploy to DigitalOcean on: push: branches: [main] pull_request: branches: [main] jobs: test: runs-on: ubuntu-latest services: postgres: image: postgres:15 env: POSTGRES_PASSWORD: postgres options: >- --health-cmd pg_isready --health-interval 10s --health-timeout 5s --health-retries 5 steps: - uses: actions/checkout@v4 - name: Set up Node.js uses: actions/setup-node@v4 with: node-version: '20' cache: 'npm' - name: Install dependencies run: npm ci - name: Run tests run: npm test env: DATABASE_URL: postgresql://postgres:postgres@localhost:5432/test deploy: needs: test runs-on: ubuntu-latest if: github.ref == 'refs/heads/main' steps: - uses: actions/checkout@v4 - name: Install doctl uses: digitalocean/action-doctl@v2 with: token: ${{ secrets.DIGITALOCEAN_TOKEN }} - name: Trigger App Platform deploy run: | APP_ID=$(doctl apps list --format ID --no-header) doctl apps create-deployment $APP_ID --wait ``` **Required GitHub Secrets**: - `DIGITALOCEAN_TOKEN`: Personal access token from DigitalOcean --- ## Monitoring and Observability ### Built-in App Platform Metrics DigitalOcean provides basic metrics out of the box: - CPU usage per service - Memory usage - Request count and latency (p50, p95, p99) - HTTP error rates (4xx, 5xx) - Active connections Access via: **Apps → Your App → Insights** ### Log Aggregation **Built-in Logs**: ```bash # View live logs via CLI doctl apps logs $APP_ID --follow # View specific component doctl apps logs $APP_ID --type run --follow ``` **Forward to External Service** (Datadog, Logtail, etc.): Add log shipping in your app: ```javascript // Node.js with Winston → Logtail import winston from 'winston'; import { Logtail } from '@logtail/node'; import { LogtailTransport } from '@logtail/winston'; const logtail = new Logtail(process.env.LOGTAIL_TOKEN); const logger = winston.createLogger({ transports: [new LogtailTransport(logtail)], }); logger.info('Application started', { service: 'web' }); ``` ### Database Monitoring DigitalOcean Managed Databases include: - Connection pool stats - Query performance insights - Replication lag monitoring - Disk usage alerts ```bash # View database metrics doctl databases metrics production-postgres ``` **Set up alerts**: ```bash # Create CPU alert doctl monitoring alert create \ --type v1/insights/droplet/cpu \ --threshold 80 \ --window 5m \ --entities production-postgres ``` ### APM Integration Integrate with Datadog, New Relic, or Sentry: ```bash # Add APM env vars to App Platform doctl apps update $APP_ID --set-env="DD_API_KEY=your-key" doctl apps update $APP_ID --set-env="DD_SERVICE=myapp" doctl apps update $APP_ID --set-env="DD_ENV=production" ``` --- ## Cost Optimization Tips ### 1. Use Reserved Database Capacity For predictable workloads, reserved capacity saves 20-30%: ```bash # Currently not available via CLI, purchase through UI # Databases → Manage → Reserved Capacity ``` ### 2. Right-Size Your Instances Start small, scale up based on metrics: ```yaml # Start here instance_size_slug: basic-xs # $5/month, 512MB RAM # Scale to this if needed instance_size_slug: professional-xs # $24/month, 1GB RAM ``` Monitor memory usage: If consistently >80%, upgrade. If <50%, downgrade. ### 3. Enable Auto-Scaling Only pay for capacity during traffic spikes: ```yaml autoscaling: min_instance_count: 2 max_instance_count: 10 metrics: cpu: percent: 75 ``` ### 4. Use Development Environments Wisely Don't run staging 24/7: ```bash # Pause staging app when not needed doctl apps update $STAGING_APP_ID --spec staging-app.yaml # In staging-app.yaml, set instance_count: 0 ``` Or use ephemeral preview environments (GitHub integration). ### 5. Optimize Database Connections Use connection pooling to reduce database cluster size: ```javascript // Bad: Each request creates new connection const client = new pg.Client(process.env.DATABASE_URL); await client.connect(); // Good: Use connection pool const pool = new pg.Pool({ connectionString: process.env.DATABASE_URL, max: 20, // Match your DB pool size }); ``` --- ## Common Gotchas and Troubleshooting ### 1. Connection Pool Exhaustion **Problem**: "remaining connection slots are reserved" errors. **Solution**: Use DigitalOcean's connection pooling: ```bash # Create pool with transaction mode doctl databases pool create production-postgres myapp-pool \ --db myapp \ --size 25 \ --mode transaction # Use pooled connection string (port 25061, not 25060) ``` ### 2. SSL/TLS Certificate Issues **Problem**: Database connection fails with SSL errors. **Solution**: Download CA certificate: ```bash # Download DigitalOcean CA cert curl -O https://raw.githubusercontent.com/digitalocean/pg_ssl_cert/main/ca-certificate.crt # Use in connection string postgresql://user:pass@host:25060/db?sslmode=require&sslrootcert=ca-certificate.crt # Or for Node.js const pool = new pg.Pool({ connectionString: process.env.DATABASE_URL, ssl: { rejectUnauthorized: true, ca: fs.readFileSync('./ca-certificate.crt').toString(), }, }); ``` ### 3. Environment Variable Naming **Problem**: Heroku uses `PORT`, DigitalOcean uses `APP_PORT`. **Solution**: Adjust app startup: ```javascript // Support both const port = process.env.PORT || process.env.APP_PORT || 8080; app.listen(port); ``` Or set `PORT` explicitly in App Platform env vars. ### 4. Build vs Runtime Commands **Problem**: Database migrations run during build, but DB isn't accessible yet. **Solution**: Use run commands, not build commands: ```yaml # WRONG build_command: npm run build && npm run migrate # RIGHT build_command: npm run build run_command: npm run migrate && npm start ``` ### 5. Regional Data Transfer Costs **Problem**: High data transfer fees if database and app are in different regions. **Solution**: Keep everything in same region + VPC: ```bash # Ensure all resources use same region --region nyc3 # For all components ``` Data transfer within same region + VPC is **free**. --- ## Real-World Migration Example ### Rails API + React Frontend + Sidekiq **Heroku Setup**: - 2× Standard-2X dynos (web) - 2× Standard-2X dynos (worker) - Standard-4 Postgres - Premium-5 Redis - **Total: $850/month** **DigitalOcean Migration**: 1. **App Platform** (2 services): ```yaml name: myapp-production services: - name: web instance_count: 2 instance_size_slug: professional-xs github: repo: myorg/myapp branch: main build_command: bundle exec rake assets:precompile run_command: bundle exec puma -C config/puma.rb - name: worker instance_count: 2 instance_size_slug: professional-xs github: repo: myorg/myapp branch: main run_command: bundle exec sidekiq ``` 2. **Managed Database**: ```bash doctl databases create prod-postgres \ --engine pg --version 15 --size db-s-4vcpu-8gb --region nyc3 ``` 3. **Managed Redis**: ```bash doctl databases create prod-redis \ --engine redis --version 7 --size db-s-2vcpu-2gb --region nyc3 ``` 4. **Spaces for ActiveStorage**: ```ruby # config/storage.yml digitalocean: service: S3 endpoint: https://nyc3.digitaloceanspaces.com access_key_id: <%= ENV['SPACES_KEY'] %> secret_access_key: <%= ENV['SPACES_SECRET'] %> region: us-east-1 bucket: myapp-production # config/environments/production.rb config.active_storage.service = :digitalocean ``` **Total DigitalOcean Cost**: $192/month **Savings**: 77% ($658/month = $7,896/year) **Migration Timeline**: - **Week 1**: Set up infrastructure, test deployments - **Week 2**: Configure database replication, test with prod data - **Week 3**: Cutover database, switch DNS, monitor - **Week 4**: Migrate Sidekiq jobs, decommission Heroku **Downtime**: 15 minutes (DNS propagation during cutover) --- ## Key Takeaways 1. **Cost Savings are Real**: 60-88% reduction in infrastructure costs for equivalent performance 2. **Migration is Incremental**: Parallel run + cutover minimizes risk 3. **Use Managed Services**: Don't self-manage databases just because you can 4. **Connection Pooling is Critical**: Avoids database scaling issues 5. **Regional Consistency Matters**: Keep resources in same region + VPC for free data transfer 6. **Terraform from Day 1**: Infrastructure as code prevents configuration drift 7. **Test with Production Data**: Run shadow environment before cutover --- ## The Bottom Line Migrating from Heroku to DigitalOcean isn't about abandoning managed services — it's about **choosing better-priced managed services**. With App Platform, you keep the developer experience (git push deployments, managed databases, zero-config SSL) while cutting costs by 60-80%. With Coolify self-hosted, you can achieve 90%+ savings for multi-app setups on a single $24/month Droplet. The migration itself takes 3-4 weeks with minimal downtime when done incrementally. For most teams spending >$200/month on Heroku, the savings justify the effort within 2-3 months. **When to migrate**: - Heroku bill >$200/month - You have 1+ engineer who can dedicate 20-30 hours over 3-4 weeks (App Platform) or 10-15 hours (Coolify) - Your app uses standard patterns (PostgreSQL, Redis, S3) - You want cost predictability **Choose App Platform if**: You want managed services, auto-scaling, and minimal ops work. **Choose Coolify if**: You're comfortable with server management and want maximum savings (90%+). **When to stay on Heroku**: - Bill <$100/month (migration effort not worth it) - You need Heroku-specific add-ons with no alternatives - Team has zero DevOps experience and no time to learn - You're pre-product-market-fit and optimizing for speed over cost **Get started**: [Sign up for DigitalOcean](https://www.jdoqocy.com/click-101674709-15836238) and receive $200 in credits to test your migration risk-free.

Tag: App Platform

Posts

Migrating from Heroku to DigitalOcean