1. OpenMetadata Server, pick the latest release from our (github releases)[https://github.com/open-metadata/OpenMetadata/releases]
2. OpenMetadata Server stores the metadata in a relational database. We support MySQL or Postgres. Any Cloud provider such as AWS or GCP's provided MySQL/Postgres services will work as well
OpenMetadata requires connectors to be scheduled to periodically fetch the metadata or you can use the OpenMetadata APIs to push the metadata as well.
1. OpenMetadata Ingestion Framework is flexible to run on any orchestrator. However we built an ability to deploy and manage connectors as pipelines from the OpenMetadata UI. This requires the Airflow container we ship. If you have an existing Airflow you can also install the REST APIs we developed for it by following the [steps here](https://docs.open-metadata.org/v1.1.1/deployment/ingestion/airflow)
2. If your team prefers to run on any other orchestrator such as prefect, dagster or even github workflows. Please refer to our recent webinar on [How Ingestion Framework works](https://www.youtube.com/watch?v=i7DhG_gZMmE&list=PLa1l-WDhLreslIS_96s_DT_KdcDyU_Itv&index=10) or refer to [docs here for an example](https://docs.open-metadata.org/v1.1.1/connectors/database/snowflake)
The Dependencies must be deployed before installing OpenMetadata Application. With Docker and Kubernetes deployment, we package these [together as part of quickstart](https://docs.open-metadata.org/v1.1.1/quick-start).
For production deployment, we recommend you to bring up MySQL/Postgres, Elastic/OpenSearch, Scheduler (like airflow) as an external service and do not rely on quickstart packages.