HeatWave User Guide
Note
These release notes were created with the assistance of MySQL HeatWave GenAI.
-
The
NL2MLroutine includes a newskip_generateparameter, which lets you to skip content generation and just return citations, enabling integration with external LLMs or MCP servers for response generation or relevant data reference.For more information, see NL2ML and Use NL2ML with In-Database LLMs. (Bug #38397642)
-
OpenAI and Gemini models are now available in MySQL HeatWave GenAI with OCI Generative AI Service LLMs.
To learn how to view the list of available models, see View Available Models. (Bug #38752998)
-
MySQL HeatWave Lakehouse now supports the loading of complex and nested Parquet types, including lists, maps, and other nested structures, which are flattened into a JSON representation. This enhancement enables you to load any complex type into a JSON column, while still maintaining existing conversions for simple lists to vectors and strings. With this update, Parquet columns of type list, map, or nested/group nodes are inferred as JSON, providing a more comprehensive and flexible data loading experience.
For more information, see Parquet File Format. (WL #17152)
-
MySQL HeatWave Lakehouse now supports load validation of external tables with the introduction of the
VALIDATE ONLYclause for theALTER TABLE ... SECONDARY_LOADstatement. This enhancement enables you to validate whether a load of a table may succeed without actually loading the data into the secondary engine, providing an additional layer of control and flexibility in managing external tables. TheVALIDATE ONLYclause can be used to validate a specified number of rows or all rows, and also supports Guided Load options, making it easier to ensure data integrity and optimize loading processes.For more information, see Validate Load. (WL #17124)
-
MySQL HeatWave now supports auto materialized view substitution, a powerful internal optimization that enhances query performance. This feature enables the MySQL HeatWave optimizer to automatically substitute a subtree of a query with an existing materialized view, leading to more efficient query execution. With this update, you can expect improved performance without any changes to the external interface, as the way to create MySQL HeatWave materialized views remains the same. This enhancement is designed to work seamlessly in the background, enabling you to benefit from faster query results without requiring any additional configuration or setup.
For more information, see Automated Substitution of Materialized Views and rapid_auto_create_materialized_view Variable. (WL #17104)
-
MySQL HeatWave now supports Grouping Sets, a powerful extension that enables you to specify multiple groupings of data in a single query. This feature provides a concise way to simplify complex queries and improve performance by reducing the number of passes over the data. With Grouping Sets, you can define one or more grouping sets, each containing a comma-separated list of group by columns or expressions, allowing for more flexible and efficient data analysis. This update enhances the capabilities of MySQL HeatWave, making it an even more robust and efficient solution for data processing and analysis.
For more information, see GROUPING SETS Modifier. (WL #16494)