Migrating from Heroku to AWS: Postgres and Beyond

Heroku changed the developer experience permanently when it launched in 2007. git push heroku main deploying a production application was genuinely magical. For over a decade, Heroku was the default answer for teams that wanted to ship fast without managing infrastructure.

The calculus shifted in November 2022 when Heroku retired its free tier and shortly after discontinued its cheapest database plans. Teams that had been staying on Heroku for cost or inertia reasons suddenly had a concrete reason to evaluate AWS. This guide is for those teams.

We are an AWS Select Tier Consulting Partner. We will be direct about what Heroku does better — and what makes AWS the stronger long-term platform for most teams.

Heroku to AWS Service Mapping

Cost Comparison

Heroku’s pricing after the 2022 changes is no longer competitive for most workloads.

Compute

Add an Application Load Balancer (~$16/month) to the AWS side — Heroku’s routing tier is included. At 2+ Dynos, AWS pulls ahead on cost.

Managed Postgres

RDS is cheaper at every tier once you account for Multi-AZ — Heroku’s HA pricing is significantly higher than RDS Multi-AZ.

The Biggest Migration Challenge: Buildpack to Docker

Heroku’s buildpack system is elegant. You push code; Heroku detects the runtime, installs dependencies, and produces a slug. No Dockerfile required.

AWS expects a container image. Converting a Heroku app to Docker is the most common stumbling block in this migration.

The process:

1. Create a Dockerfile that replicates what your buildpack does: install your language runtime, copy your code, install dependencies, and set an entrypoint.
2. Test the container locally with docker build and docker run, passing environment variables to simulate Heroku Config Vars.
3. Push the image to Amazon ECR (Elastic Container Registry).
4. Reference the ECR image in your ECS task definition or Lambda container image.

For Rails, Django, Node.js, and other common frameworks, community-maintained Docker base images make this straightforward. The heroku/heroku Docker base images are also an option for a more direct lift-and-shift. For exotic buildpacks or multi-buildpack stacks, expect additional debugging time.

Migrating Heroku Postgres to RDS

Option 1: pg_dump / pg_restore (Small to Medium Databases)

# Export from Heroku
heroku pg:backups:capture --app your-app-name
heroku pg:backups:download --app your-app-name

# Restore to RDS
pg_restore --verbose --clean --no-acl --no-owner \
  -h your-rds-endpoint.rds.amazonaws.com \
  -U postgres -d your_database_name \
  latest.dump

This approach requires downtime equal to your dump-and-restore time. For databases under ~10 GB, expect 15–60 minutes.

Option 2: AWS DMS (Large Databases or Near-Zero Downtime)

AWS Database Migration Service treats Heroku Postgres as a standard PostgreSQL source. Configure DMS with a full-load task followed by ongoing change data capture (CDC) replication. This keeps RDS in sync with Heroku while you validate your application. Cutover is a connection string change and a DMS task stop.

Post-Migration Steps

• Update your environment variables (Heroku Config Vars → AWS Secrets Manager or Parameter Store) to point at the new RDS endpoint.
• Remove the ssl=true Heroku Postgres certificate workaround if present; RDS uses standard SSL certificates.
• Test all database-heavy application paths before cutting over production traffic.

CI/CD Re-plumbing: Heroku Auto-Deploy → GitHub Actions

Heroku’s GitHub auto-deploy integration is one of its best features — push to main, get a deployment. AWS does not have an equivalent single-click integration, but GitHub Actions fills the gap effectively.

A minimal pipeline for an ECS Fargate deployment:

1. On push to main, GitHub Actions builds your Docker image and pushes it to ECR.
2. A subsequent job updates the ECS service with the new task definition, triggering a rolling deployment.

This requires ~60 lines of YAML versus Heroku’s zero-line integration. The trade-off is control — you can add test gates, approval steps, multi-environment promotion, and deployment notifications.

For teams migrating from Heroku Pipelines (staging → production promotion), the AWS equivalent is CodePipeline with ECR and ECS deploy stages, or a multi-environment GitHub Actions workflow with manual approval gates.

What Heroku Does Better

Heroku’s advantages are real and worth naming:

• Developer experience. git push heroku main remains unmatched for speed. AWS deployment always requires more configuration.
• Zero DevOps required. No VPC, no security groups, no IAM roles. For a 2-person team shipping fast, this matters.
• Review Apps. Heroku creates per-PR preview environments automatically. AWS has no equivalent native feature — approximating this with ECS or Lambda requires custom tooling.
• Add-on ecosystem. One-click Postgres, Redis, SendGrid, Papertrail, and hundreds of others. AWS services require more configuration.

When AWS Wins

• Cost at scale. Beyond 2–3 Standard Dynos and a Standard-0 Postgres, AWS is materially cheaper.
• Compliance. HIPAA, PCI DSS, and FedRAMP are not available on Heroku.
• AWS service integration. If your app needs SES, SQS, Bedrock, or any other AWS service, running on AWS itself reduces latency and eliminates cross-cloud egress charges.
• Infrastructure control. VPC isolation, custom networking, private subnets, and security groups are unavailable on Heroku.
• Data residency. AWS provides 33 regions globally; Heroku is limited to a handful of regions.

Ready to Migrate?

FactualMinds handles Heroku-to-AWS migrations — from Postgres migration and Docker containerization to ECS Fargate setup and CI/CD pipeline re-plumbing.

Talk to our team about your Heroku migration or learn more about our AWS Migration service.