Oracle released new generative AI features among other updates to its managed data analytics and database service, Oracle HeatWave at its CloudWorld 2024 conference.
Oracle HeatWave encompasses several modules including HeatWave Lakehouse, HeatWave on AWS, HeatWave AutoML, HeatWave Gen AI, and MySQL HeatWave.
The generative AI-based updates have been added to the Gen AI module and these include multi-lingual support, optical character recognition (OCR) support, LLM inference batch processing, JavaScript support, and automatic vector store updates, the company said, adding that the module allows developers to build generative AI-based application without data movement or additional cost.
The multi-lingual support, according to the company, will allow developers to build global applications as it supports loading documents in 27 languages into HeatWave’s Vector Store to perform similarity searches and ask questions in various languages.
The OCR support is designed to allow developers to enable similarity searches for end users by leveraging the Vector Store to convert scanned content saved as images into text data, which can also be analyzed.
The LLM inference batch processing enhancement is designed to help developers improve application throughput by executing multiple requests simultaneously across the HeatWave cluster.
In June, Oracle became the first database vendor to offer added support for LLMs inside a database — a move targeted at reducing costs.
Other updates to the module include JavaScript support and automatic vector store updates, which are expected to allow developers to use JavaScript with the Vector datatype, invoke HeatWave GenAI from a JavaScript program, and build applications using the latest data, respectively.
“Any changes to documents in object storage automatically trigger updates to corresponding vector embeddings,” the company said, explaining the automatic vector store update feature.
HeatWave MySQL gets cost and performance enhancements
HeatWave MySQL, formerly known as MySQL HeatWave, has been updated with three key features designed to reduce costs and improve the performance of the cloud-based analytics module.
The key updates include a hypergraph optimizer, integration with OCI Ops Insights, and bulk ingest.
While the hypergraph optimizer enables users to achieve true cost-based join optimization of query plans and improve performance, particularly for complex queries, the integration with OCI Ops Insights helps administrators uncover performance issues, forecast consumption, and plan capacity using ML-based analytics.
The bulk ingest feature, on the other hand, enables users to load data into HeatWave MySQL up to five times faster, resulting in data being queried sooner, and resources being freed up faster, reducing costs, the company explained.
HeatWave AutoML gets larger ML model training capability
The machine learning module inside HeatWave has been updated with new capabilities that make it easier for developers to train machine learning (ML) models, the company said.
The new capabilities include the ability to store and process larger ML models, topic modeling, data drift, and semi-supervised log anomaly detection.
While the ability to store and process larger ML models enables users to train ML models with a richer training data set by increasing capacity to accommodate four times larger models, the topic modeling capability enables users to discover insights in large textual data sets by helping them understand key themes in documents, the company explained.
The data drift feature, on the other hand, helps users determine when to retain models by detecting the differences between the data used for training and new incoming data.
Separately, the semi-supervised log anomaly detection capability allows users to provide feedback on the results of unsupervised anomaly detection and use this labeled data to help improve subsequent predictions, the company said.
HeatWave Lakehouse gets updated object storage features
The data lakehouse module inside HeatWave, according to the company, has been updated with new capabilities, such as support for writing results to object storage and automatic change propagation.
While the support for writing results to object storage enables users to easily share query results and store them in object storage inexpensively, the automatic change propagation feature enables users to always query the latest data by automatically detecting data changes in object storage and updating those changes incrementally to HeatWave, the company explained.
HeatWave has been made available as part of the OCI Always Free Service, which enables enterprises to develop and run small-scale applications using HeatWave MySQL, analytics, machine learning, JaveScript, HeatWave Vector Store, and process data in object store.
“All OCI accounts get access to a standalone HeatWave instance in OCI in their home region, along with 50 GB of storage and 50 GB of backup storage, for an unlimited time. They also receive $300 of credit to use for all eligible OCI services for up to 30 days,” the company explained.
The updates made to HeatWave running on Oracle Cloud Infrastructure (OCI) have also been added to HeatWave on AWS. These features and capabilities include generative AI updates, updates to Lakehouse, native JavaScript support, and autopilot indexing.