1. Lakehouse Basics

A lakehouse combines low-cost, scalable lake storage and open data formats with table metadata, governance, and SQL analytics. Oracle AI Database can query supported external files and catalog-managed tables in place, join them with database-resident data, and, for supported Iceberg configurations, create and write tables in lake storage.

This guide distinguishes three primary concepts of Oracle AI Database Lakehouse:

Concept	What it Defines	Examples
File Format	How rows and columns are encoded inside a file.	Parquet, ORC, Avro, CSV, JSON, text
Table Format	How a logical table is represented across data files and metadata over time.	Apache Iceberg
Catalog	How named tables and namespaces resolve to current metadata.	Oracle AI Data Catalog, AWS Glue, Snowflake Polaris, Databricks Unity Catalog
Sharing Protocol	How a provider grants a consumer controlled access to shared data.	Delta Sharing