Apache Iceberg for Beginners: Build a Lakehouse Anywhere ft. MinIO

Summary

Apache Iceberg explained for beginners: what a data lakehouse is, how Iceberg tables and the Iceberg REST catalog work, and how to run a high-performance lakehouse on cloud, on-prem, or hybrid infrastructure. In this OLake Community Call, Brenna Buuck, Developer Evangelist at MinIO and a data engineer for 11 years, covers the foundations of building Apache Iceberg data platforms, based on her work with teams running Iceberg lakehouses in production.

• The three layers of a data lakehouse: object storage, an open table format and a query engine, each decoupled from the others
• How a lakehouse differs from a data warehouse and a data lake
• How Apache Iceberg handles schema evolution, hidden partitioning and engine independence
• How Iceberg metadata files, manifest lists and manifests let query engines skip files
• Why object storage sits on the performance-critical path for Iceberg queries
• How the Iceberg REST catalog tracks snapshots, enables time travel and handles concurrent writes
• What's new in Iceberg v3: deletion vectors, row lineage, variant types and geospatial types
• Three ways to run an Iceberg catalog: standalone, embedded in object storage and federated