OpenMetadata 627,669 (https://github.com/open-metadata/OpenMetadata 15,365 ) (Apache-2.0, about 15,400 stars) was open-sourced in 2021 by Suresh Srinivas and Sriharsha Chintalapani, who had built Uber's Databook catalog; their company Collate backs it and sells a hosted edition. Version 2.0.3 shipped on 30 September 2026. It is smaller than DataHub 559,249 : one Java server, MySQL 524 or PostgreSQL 1,289 , and Elasticsearch or OpenSearch, with webhooks instead of Kafka 129 . Ingestion runs as Python workflows (pip 21,050 install openmetadata-ingestion) under Airflow 129 or any orchestrator.
Its standout is built-in auto-classification: column-name patterns plus an NLP recognizer over sampled values tag PII.Sensitive above a confidence threshold (80 by default), which is PII Classification and Masking as a product feature. It also runs data-quality tests from its UI and puts glossary terms through an approval workflow.
Not run here: its quick start asks Docker 514 for at least 6 GiB and 4 vCPUs, this whole shared host. Choose it for catalog, quality and governance in one product with fewer moving parts; choose DataHub when metadata must flow as events into other systems or you already run Kafka.