Oracle has introduced new generative AI features and several other enhancements to its managed data analytics and database service, Oracle HeatWave, at CloudWorld 2024. The updates span across multiple modules, including HeatWave Gen AI, MySQL HeatWave, HeatWave Lakehouse, and HeatWave AutoML.
Key generative AI updates to the HeatWave Gen AI module include:
- Multi-lingual support for 27 languages to enable global applications with multi-language document handling.
- Optical Character Recognition (OCR) for converting scanned image content into text data for analysis and similarity searches.
- LLM inference batch processing to improve throughput by handling multiple requests simultaneously.
- JavaScript support and automatic vector store updates for enhanced development flexibility and real-time data synchronization.
Oracle has also introduced performance and cost enhancements to HeatWave MySQL, including a hypergraph optimizer for cost-based join optimization, integration with OCI Ops Insights for capacity planning, and a bulk ingest feature to speed up data loading by five times.
The HeatWave AutoML module has been updated with capabilities to store and process larger ML models, topic modeling for large text datasets, data drift detection, and semi-supervised log anomaly detection for improved predictions.
Lastly, HeatWave Lakehouse received updates that include support for writing query results to object storage and automatic change propagation to ensure real-time querying of updated data. Oracle continues to extend HeatWave’s availability across Oracle Cloud Infrastructure (OCI) and AWS, with features like generative AI updates, JavaScript support, and autopilot indexing available on both platforms.
