How to Use This Guide
This guide is for database architects, database administrators, data engineers, analytics engineers, and application teams that plan or operate Oracle AI Database lakehouse access. Use it to choose an access path, establish the required setup, and operate the path within its documented deployment and release boundaries.
Guide Navigation
| Chapter | Focus Area |
|---|---|
| 1. Lakehouse Basics | Architecture, formats, metadata, access patterns, and product boundaries. |
| 2. Setup and Access | Deployment checks, network access, identities, credentials, least privilege, and diagnosis. |
| 3. Catalogs, Governance, and Enterprise Data Access | AI Catalog, source catalogs, per-database mounts, metadata ownership, and governance. |
| 4. Working with Iceberg Tables | Read existing Iceberg tables and use the Oracle-managed create/write path through one supported mounted-catalog lifecycle. |
| 5. External Files and External Tables | Persistent, inline, and catalog-mounted SQL access to externally managed data. |
| 6. Performance Considerations | Pruning, statistics, parallelism, Iceberg metadata, Lake Cache, and materialization decisions. |
Use the Guide in this Order
- Identify the target Oracle deployment, database release update, catalog provider, storage provider, and intended operation.
- Use the deployment and capability matrix below to determine which documentation set controls support for that path.
- Complete Chapter 2 before creating credentials, mounts, or external objects.
- Use Chapters 3 through 5 for the selected data-access pattern, then use Chapter 6 to diagnose and optimize the workload.
Scope and Conventions
- Supported means explicitly documented for the selected deployment and release update. This guide does not establish support for an unlisted provider, operation, or feature.
- Lakehouse is an architecture that combines open data storage, table metadata, governance, and query engines. AI Lakehouse is an Oracle deployment and product term; it is not a synonym for the architecture.
- Package availability, network configuration, cloud-native identities, catalog integrations, Lake Cache, and Data Lake Accelerator can vary by deployment.
Deployment and Capability Matrix
| Deployment | Capability variables | Verify before using a path |
|---|---|---|
| Customer-managed Oracle AI Database 26ai | Installed packages, outbound network and TLS configuration, local ACLs, provider integration, and release update. | Confirm the documented database release update, required packages, and supported catalog, storage, and authentication combination. |
| Exadata Database Service and Exadata Cloud@Customer | Service configuration, network topology, package availability, and provider integration. | Use the service documentation for the selected deployment and release update to qualify the capability. |
| AI Lakehouse | Managed-service integrations, deployment-provided identities, and availability of Lake Cache or Data Lake Accelerator. | Use the service documentation for the selected region and release update before relying on a managed capability. |