Deployment Overview

Causet is a suite of services deployed together. This section covers how to run them in production, what infrastructure they depend on, and how to move from a compiled IR artifact to a live deployment.

Services

Service	Language	Port(s)	Role
`causet-runtime-service`	Java / Spring WebFlux	8080 (HTTP), 9090 (gRPC)	Intent processing, ledger writes, rule evaluation
`causet-query-service`	Java / Spring WebFlux	8082	Named query execution against projection tables
`causet-projection-worker`	Java / Spring WebFlux	— (Kafka consumer)	Consumes projection events, UPSERTs into projection tables
`causet-saas-cloud`	Java / Spring	8085	Control plane API: platforms, applications, releases, forks
`causet-cloud-control-plane`	Next.js	3000	Web UI for the control plane
`ws-gateway-go`	Go	8081	WebSocket gateway for real-time event push
`causet-compiler`	Java	— (CLI)	Compiles `.causet` DSL files into IR artifacts
`causet-cli`	Go	— (CLI)	Developer CLI: deploy, release, query

Service dependency graph

Required infrastructure

Component	Purpose	Local equivalent
PostgreSQL (causet DB)	Event store (`ledger_events`) + tenant schemas	Docker Compose Postgres
PostgreSQL (projections DB)	Projection tables per tenant	Same Postgres instance, separate DB
Kafka / Redpanda	Event streaming between runtime and projection worker	Redpanda in Docker Compose
Redis	IR artifact cache + optional query result cache	Redis in Docker Compose
S3 / MinIO	Compiled IR artifacts per release version	MinIO in Docker Compose
Clerk	JWT authentication for control plane and SaaS API	Clerk dev instance

Deploy artifacts

A Causet release consists of two IR artifacts produced by causet-compiler:

causet.runtime.json — rule definitions, action schemas, event definitions loaded by causet-runtime-service
causet.projections.json — projection definitions and DDL loaded by causet-projection-worker and causet-query-service

Both artifacts are uploaded to S3 under a versioned path:

s3://{bucket}/{irVersion}/causet.runtime.json
s3://{bucket}/{irVersion}/causet.projections.json

Services fetch their IR artifact from S3 on startup and when a new release is activated.

Release flow

Compile — run causet-compiler against your .causet files
Upload IR — push both artifacts to S3 at the versioned path
Create release — register the IR version in causet-saas-cloud
Deploy to fork — activate the release on a specific fork (e.g., production)
Services reload — runtime and projection worker fetch new IR, DDL is applied

Deployment models

Local Docker Compose

All services run in a single docker compose up -d. Infrastructure (Postgres, Redpanda, Redis, MinIO) is included. Suitable for development and integration testing.

See Local Stack.

AWS ECS

Production deployment uses AWS ECS with:

RDS PostgreSQL (two instances or two databases)
Amazon MSK or self-managed Redpanda on EC2
ElastiCache Redis
S3 for IR artifacts
ECR for Docker images
ALB for HTTP routing

See AWS / ECS.

Any container platform

Causet services are standard Docker containers. They can run on Kubernetes, Fly.io, Railway, or any platform that supports Docker. The only requirements are network connectivity to the infrastructure listed above and the environment variables described in Environment Variables.

Section pages

Local Stack — docker compose up for local development
Docker — building images, container configuration
AWS / ECS — production deployment on AWS
Database Migrations — how schema changes work
Environment Variables — complete env var reference
Health Checks — /actuator/health and ECS configuration
Release Checklist — pre-deploy and post-deploy verification

What is Causet?