Here are the key Prefect 3.x docs for scaling a self‑hosted (OSS) server: - Scale self-hosted Prefect – End-to-end guide for multi-server deployments (API replicas, background services, Redis, Postgres, LB) - Server concepts – Server architecture, components, and how they interact - Database maintenance – Postgres sizing, retention, and maintenance (critical at scale) Deployment options: - Self-host with Docker Compose – Reference compose and env config - Self-host with Helm (Kubernetes) – Production-ready K8s patterns Helm resources: - Prefect Helm charts repo - Server chart values - Worker chart values Workers and execution: - Work pools – Configure your execution backends - Workers – Scaling workers and concurrency limits Given you’re on ECS/EC2: a common production pattern is RDS (Postgres) + ElastiCache (Redis), multiple

prefect server start --no-services

API tasks behind an ALB, and a single

prefect server services start

task for background services. If you want, I can share a reference ECS task/service layout and health check config tailored to your setup.