Rivet Engine
What is Rivet Engine?
Rivet Engine is the backend engine that powers Rivet Actors at scale. It's an optional component of the Rivet ecosystem that provides enterprise-grade infrastructure for running stateful workloads with long-lived processes, durable state, and realtime capabilities.
While you can use RivetKit (the TypeScript library) with a file system or memory driver for development, Rivet Engine provides production-ready infrastructure for:
- Automatic scaling from zero to millions of concurrent actors
- Distributed state management with fast, in-memory state storage
- Actor lifecycle management with hibernation and instant wake-up
- Built-in resilience with automatic failover and recovery
- Edge deployment for low-latency global distribution
Think of it as the production runtime that takes your RivetKit actors and scales them effortlessly across distributed infrastructure.
Key Features
- Production-Ready Scaling: Automatically scale from zero to millions of concurrent actors with no cold starts
- Distributed State Storage: State is stored on the same machine as your compute for ultra-fast reads and writes with no database latency
- Actor Lifecycle Management: Actors automatically hibernate when idle and wake instantly on demand, only consuming resources when active
- Built-In Realtime: Native support for WebSockets and SSE for realtime state updates and broadcasting
- Resilient Infrastructure: Automatic failover and recovery with state integrity preservation
- Edge-Ready: Deploy actors close to users for instant interactions
- Multi-Runtime Support: Works with Node.js, Bun, Deno, and Cloudflare Workers
- API Integration: RESTful API for managing runners, actors, and namespaces
- Built on Web Standards: Uses standard HTTP, WebSockets, and SSE protocols
Use Cases
Rivet Engine powers applications that require:
- AI Agents: Durable AI assistants with persistent memory and realtime streaming capabilities
- Realtime Collaboration: Collaborative documents, whiteboards, and tools with CRDTs and realtime synchronization
- Durable Workflows: Multi-step workflows with automatic state management and recovery
- Local-First Applications: Offline-first apps with server-side synchronization and conflict resolution
- Chatbots & Automation: Discord, Slack, or autonomous bots with persistent conversation state
- Per-User Databases: Isolated data stores for each user with zero-latency access
- Multiplayer Games: Authoritative game servers with realtime state synchronization
- Background Processing: Scheduled and recurring jobs without external queue infrastructure
- Rate Limiting: Distributed rate limiting with in-memory counters and state
How It Works
Rivet Engine provides a backend runtime that:
- Manages Actors: Coordinates actor lifecycle, hibernation, and wake-up across distributed infrastructure
- Handles State: Stores actor state in memory for fast access, with persistent backup to storage backends
- Routes Requests: Efficiently routes client requests to the correct actor instance
- Scales Dynamically: Automatically spawns and destroys actor instances based on demand
- Ensures Resilience: Monitors health and automatically recovers failed actors with state preservation
Integration with RivetKit
Rivet Engine works seamlessly with RivetKit, the TypeScript library for building actors:
- Development: Use RivetKit with file system or memory drivers for local development
- Production: Deploy to Rivet Engine for automatic scaling and enterprise features
- Self-Hosted: Run Rivet Engine on your own infrastructure for full control
- Managed: Use Rivet Cloud for 1-click deployment with managed infrastructure
Deployment Options
Deploy Rivet Engine on:
- Self-Hosted (Moltern, Docker, Kubernetes)
- Cloud Platforms (Vercel, Railway, AWS ECS, Google Cloud Run, Hetzner)
- Rivet Cloud (Managed, 1-click deployment)
Storage Backends
Rivet Engine supports multiple storage backends for actor state:
- PostgreSQL
- File System
- Memory (development/testing)
- Custom storage adapters