These release notes were created with the assistance of HeatWave GenAI.
-
Context search in HeatWave GenAI now lets you use your own embedding tables. This enhancement enables more flexible query enrichment for Large Language Models (LLMs). This update extends the search capabilities of the
ML_RAG
,ML_RAG_TABLE
, andHEATWAVE_CHAT
routines to include your own embedding tables.For more information, see Using Your Own Embeddings With Retrieval-Augmented Generation and Running HeatWave Chat. (WL #16537)
-
HeatWave now supports automatic recovery of loaded InnoDB tables from Storage Layer after a planned or unplanned restart of the primary DB system. This feature extends the existing functionality of automatic recovery from Storage Layer for Standalone DB systems to HA DB systems, thus reducing the recovery time, increasing the availability, and improving the overall system reliability. However, Lakehouse tables in HA DB systems still need to be recovered from Object Storage.
For more information, see Scale-Out Data Management. (WL #16627)
-
HeatWave now supports automated zone map construction on non-primary key columns, improving RAPID query processing performance. This feature uses a workload-driven approach to select the most beneficial zone map columns based on query history, continuously adapting to the workload.
For more information, see Zone Maps. (WL #15993)
-
The
POOL_TYPE
column in HeatWave Performance Schema tables no longer includes theRAPID_LOAD_POOL_
prefix. The possible values are nowSNAPSHOT
andTRANSACTIONAL
.For more information, see The rpd_tables Table. (WL #16630)
-
For tracking the usage of various components, HeatWave Option tracking now supports a single, fast counter,
usedCounter
, instead of a boolean value,used
. The counter indicates the number of times a component is used.For more information, see Option Tracker. (WL #16721, WL #16732)