Herding Cat(alogue)s: The Apache Iceberg™ Catalog Landscape
summary
Concrete comparison of Iceberg catalogs: HMS uses database locks and Thrift so it’s ubiquitous but coarse‑grained; AWS Glue uses conditional writes and a proprietary API with managed credential vending but costs scale with API use; JDBC is fine for dev only; Nessie offers git‑like branching; Gravitino and Polaris provide federation and REST‑native servers that add multi‑table transactions and server‑side conflict resolution; portability and ecosystem fit drive the choice, so prefer a REST catalog spec implementation for long‑term flexibility and credential vending.
AI-written summary · about
chapters (14)
- 0:00Introduction and overview
- 1:54Data lakehouse components
- 4:02Iceberg table format basics
- 7:52Role and need for a catalog
- 10:01REST catalog specification
- 15:14Hive Metastore overview
- 17:21AWS Glue catalog
- 19:47Apache Polaris introduction
- 23:17Databricks Unity Catalog
- 25:44Other catalog options overview
- 27:52Comparison of main catalogs
- 29:21Choosing the right catalog
- 33:12Future of Iceberg catalogs
- 34:16Q&A session
related talks
34mApache Iceberg.
39mMetadata as a Control Plane: Inside the Apache Iceberg REST Catalog
31mOpen Standards and Lakehouse AI Development
37mApache Iceberg: An Architectural Look Under the Covers
01h 12mFrom Storage Formats to Open Governance: The Evolution to Apache Polaris
39mIntroducing Apache Iceberg: Tables Designed for Object Stores
46mTake the Polar Plunge: A Fearless Introduction to Apache Iceberg™
52mApache Iceberg - An Architectural Look Under the Covers