MySQL HeatWave AutoML now supports enhanced data preparation capabilities with NL2ML, enabling you to efficiently prepare your data for training and testing. This update introduces additional data preparation information to NL2ML's responses that enable you to join dataset tables, exclude unnecessary columns, and split your data into training and testing tables through the use of a new TRAIN_TEST_SPLIT procedure. With these new capabilities, you can streamline your data preparation workflow, making it easier to get started with machine learning tasks.

For more information, see Prepare Training and Testing Datasets. (WL #17057)

MySQL HeatWave AutoML now supports enhanced machine learning capabilities with improved performance and robustness in the ML_PREDICT_TABLE and ML_EXPLAIN_TABLE routines. The new flow loads the entire input table into the ML driver, enabling more ML operations to run in parallel, resulting in better overall performance. Additionally, the update ensures that large input tables are handled efficiently by returning results in blocks, preventing failures due to the max_allowed_packet size limitations. Furthermore, to maintain consistency in log anomaly detection tasks, the input table now requires a primary key, enhancing the reliability of the results.

For more information, see Generate Predictions for a Table and Generate Prediction Explanations for a Table. (WL #16905)

MySQL HeatWave AutoML and MySQL HeatWave GenAI now support enhanced concurrency, enabling the processing of multiple queries simultaneously across different sessions, which significantly improves overall performance and efficiency. This update allows for a more seamless and responsive experience, making it easier for you to leverage the full potential of MySQL HeatWave AutoML and MySQL HeatWave GenAI for your machine learning and generative AI workloads.

For more information, see System Variables. (WL #16847)

MySQL HeatWave GenAI now lets you generate SQL queries from natural-language statements using the new NL_SQL routine, making it easier for you to interact with databases. This feature collects information on the schemas, tables, and columns that you have access to, and then uses a Large Language Model (LLM) to generate an SQL query for the question pertaining to your data. It also lets you run the generated query and view the result set.

For more information, see Generate SQL Queries From Natural-Language Statements. (WL #17005, WL #16315)

MySQL HeatWave GenAI now supports using Oracle Cloud Infrastructure (OCI) Generative AI Service embedding models during vector store ingest using Auto Parallel Load, enabling you to create vector stores using advanced models and improving the accuracy and relevance of results.

For more information, see Ingest Files Using Auto Parallel Load. (WL #17060)

MySQL HeatWave GenAI now supports Vision Language Models (VLMs) provided by OCI Generative AI Service, enhancing its capabilities and expanding the range of available AI-powered features. This update provides you with more advanced and efficient processing of vision-language tasks within MySQL HeatWave GenAI.

For more information, see OCI Generative AI Service LLMs, ML_GENERATE, and ML_GENERATE_TABLE. (WL #17042)

MySQL HeatWave GenAI now supports advanced document summarization, enabling you to effectively summarize unstructured documents stored in Object Storage.

For more information, see Summarize Unstructured Files. (WL #16545)

MySQL HeatWave Lakehouse now supports event-based automatic incremental load on OCI, allowing for seamless and automated data refresh of tables loaded from object storage. With this update, customers can configure a Lakehouse table to enable automated incremental loading of object changes into corresponding Lakehouse tables. This feature provides an efficient and automated way to keep data up-to-date, eliminating the need for manual refreshes. This feature is available for all three HeatWave node shapes: Standard, Small, and Free.

For more information, see Refresh Data Using Event-Based Incremental Load. (WL #16513)

MySQL HeatWave Lakehouse now supports an optional clause for the ALTER TABLE SECONDARY_LOAD statement, enabling you to control whether MySQL HeatWave Guided Load is used or not. This enhancement provides flexibility and choice, enabling you to opt-out of Guided Load when loading tables, if desired. With this update, you can specify the GUIDED {ON | OFF} clause to determine whether Guided Load should be performed, making it easier to manage your data loading processes.

For more information, see Load Tables and Load Data from Object Storage. (WL #17062)

MySQL HeatWave now supports improved query performance through the introduction of a statistics cache from the InnoDB execution engine. This enhancement leverages statistics feedback to optimize query plans, particularly for queries with multiple joins and high MySQL costs. By injecting actual cardinalities from previous query executions into subsequent optimizations, MySQL HeatWave can produce more accurate and efficient plans, leading to significant performance improvements. With this update, MySQL HeatWave delivers enhanced query performance and more efficient use of system resources.

This feature can be controlled through session and global variables rapid_enable_my_sc and rapid_ap_stats_cache_max_entries.

For more information, see System Variables. (WL #16701)