Skip to main content

Self-Host

Choose self-hosting, BYOW (Bring Your Own Workers), or fully managed BYOC (Bring Your Own Cloud).

Self-host vs BYOW vs BYOC

Choose where your control plane and workers run, and who manages them.

Self-hostBYOWBYOC
Control plane and storageYour infrastructureRivet CloudYour VPC
Managed byYouRivetRivet
WorkersYou manageYou manageYou manage

Self-host

You run and maintain the workers, control plane, and storage. Choose this when you need to operate the stack yourself, including fully air-gapped deployments.

YOUR INFRASTRUCTUREWorker(your code)Control planeStorage

BYOW (Bring Your Own Workers)

You run your workers on your infrastructure. Rivet Cloud hosts and manages the control plane and storage, so you only maintain your application code and worker infrastructure.

YOUR INFRASTRUCTURERIVET CLOUDWorker(your code)Control planeStorage

See worker deployment options and manage your project in the Rivet dashboard.

BYOC (Bring Your Own Cloud)

Rivet Enterprise Edition and FoundationDB run inside your VPC, fully managed by Rivet. An operator in your cluster pulls deployment instructions from Rivet Cloud and maintains the deployment. You manage your workers.

Bring your own cloud, managed by Rivet Your VPC contains your workers, the Rivet operator, control plane and FoundationDB storage. The operator pulls commands from Rivet Cloud and manages the control plane and storage. Workers connect to the control plane inside your VPC. YOUR VPC Rivet Cloud Rivet operator Pull commands Updates & maintenance Worker (your code) Control plane Storage FoundationDB Fully managed by Rivet

See the BYOC quickstart.

Architecture

Rivet has 3 core components:

  • Worker: A process running your code with the Rivet SDK. It connects outbound to the control plane, so it never needs a public URL or an inbound firewall rule.
  • Control plane: The orchestration service that manages actor lifecycle, routes messages, and serves the API.
  • Storage: The persistence layer for actor state and the messaging infrastructure for real-time communication.

Storage backends

  • File system: Recommended for single-node deployments
  • PostgreSQL: Recommended for multi-node deployments; production-ready for light-to-moderate workloads, not built for enterprise scale
  • FoundationDB: Best scalability and performance for production deployments (requires enterprise)

See Storage for configuration.

Next steps