Description:
MySQL 9.7 LTS has support for VECTOR data type and a column in an InnoDB table can be of VECTOR type. However, it remains write-only metadata i.e. it is excluded from all index types. Core similarity primitives (for example DISTANCE()) are not implemented/exposed in community version.
Goal of this feature request is to make MySQL fully functional VECTOR store with VECTOR INDEX capabilities. Which may include
- Make MySQL a native VECTOR store with VECTOR INDEX support.
- Provide native, ACID-compliant storage for vector data.
- Provide vector distance/similarity functions (DOT, COSINE, EUCLIDEAN).
- Provide native vector query execution for similarity search on multi-dimensional vectors.
- Provide ANN indexing and search, with algorithm (eg : HNSW) for low-latency retrieval.
With this, users get one system for both transactional app data and AI retrieval, instead of stitching MySQL + external vector DB.
- Better consistency: vectors and source rows commit/rollback together.
- Stronger reliability: ACID + crash recovery for vector data.
- Easier governance: existing MySQL access control, backup, replication, and compliance processes apply to vector workloads.
In short, this work makes AI-native features practical for mainstream MySQL users without forcing additional/external db/tool.
How to repeat:
NA
Suggested fix:
NA
© 2026, Oracle Corporation and/or its affiliates