RDS for Oracle Adds 26ai, Touting AI/ML Features

RDS for Oracle just shipped 26ai. The AI features are the headline, but the JSON Relational Duality is the real win.

What changed

• Amazon RDS for Oracle now supports Oracle Database 26ai (Enterprise Edition only).
• Integration with Amazon Bedrock provides access to foundation models.
• Oracle's Select AI feature allows natural language SQL query generation.
• AI Vector Search is available for in-database vector embeddings and hybrid searches.

Why it matters

This release brings Oracle's latest Long Term Support release, 26ai, to RDS, which is significant for organizations committed to the Oracle ecosystem. The headline features are the integrated AI/ML capabilities, particularly Oracle's Select AI for natural language querying and AI Vector Search for in-database vector storage. The honest version: these features aim to reduce the operational burden of managing separate vector databases and simplify data interaction for less technical users. For existing RDS for Oracle customers, this offers a path to modernize workloads without necessarily migrating off-platform, potentially simplifying architectures by consolidating relational data, vector embeddings, and graph analytics within a single RDS instance. This could be particularly beneficial for applications requiring real-time AI-driven insights directly from transactional data.

The catch

Watch out: While the announcement touts AI/ML integration, the specifics of performance and cost for these new features are not detailed. The requirement for Oracle Database 26ai Enterprise Edition means this is not a cost-effective upgrade for all Oracle RDS users. Furthermore, migrating from a non-CDB Oracle 19c instance to 26ai requires an intermediate conversion to a CDB architecture, adding complexity and downtime. The catch: the announcement does not specify the version of Amazon Bedrock models available or any associated API call costs beyond standard RDS pricing.

Ship it

If you are running Oracle Database 19c or 21c in RDS and are evaluating Oracle 26ai's AI Vector Search or JSON Relational Duality features, plan your upgrade path. For non-CDB 19c users, factor in the conversion to CDB architecture into your migration timeline. Evaluate the potential for consolidating vector data stores with your existing relational data in RDS, potentially simplifying operations and reducing costs compared to managing a separate vector database service like Amazon OpenSearch Service or Aurora PostgreSQL with pgvector.

Bottom line: RDS for Oracle 26ai adds native AI/ML and JSON Relational Duality, but requires Enterprise Edition and careful migration planning.

— Filed to /aws