Astro

Astro is Astronomer's managed Apache Airflow service, a data orchestration platform for teams writing pipelines in Python, billed by usage per deployment hour.

Astro runs Apache Airflow for data teams that would rather write pipelines than operate the scheduler, workers and clusters underneath them. Astronomer, Inc., based in New York, builds and operates the service. Teams keep their Dags as Python files in Git, and Astro hosts the environment that code deploys into. On top of Airflow it adds observability, alerting, support and access controls. A deployment is the unit Astro manages: a hosted Airflow environment that can hibernate when idle and roll back to an earlier version. Workers scale to zero between runs, and Dedicated clusters give deployments single-tenant network isolation. The Astro CLI starts the same project on a laptop with astro dev start, and the GitHub integration deploys code when a pull request merges into a mapped branch. Preview Deployments create a separate environment for each feature branch. Astro Observe draws a lineage graph of Dags, tasks, datasets and tables and tracks freshness SLAs, while Remote Execution runs tasks on a customer's own Kubernetes. Astro fits data engineering teams already running Airflow in production who want hosting, support and monitoring without maintaining the cluster themselves. Platform teams facing security reviews get SSO enforcement, SCIM provisioning, custom RBAC, IP access lists and audit logs. Teams whose tasks must stay inside their own infrastructure keep execution there through Remote Execution while Astro hosts orchestration. Organizations that cannot accept a regional outage can add cross-region disaster recovery. Typical work on Astro is ETL and ELT, MLOps pipelines, analytics refreshes, AI workflows and moving jobs off a legacy scheduler. Through Cosmos, Astronomer's open-source package, a dbt project runs as Airflow tasks on the same schedule as the loads feeding it. Code travels from a branch to production through GitHub Actions or GitLab CI, and AI-assisted authoring helps draft and debug Dags along the way. When a run fails, Astro posts to Slack or PagerDuty with a link back to the run, so the on-call engineer can open the failed task's logs and retry history directly.