You're viewing the current version of Migration Assistant documentation (Kubernetes/EKS-based). For the classic ECS/CDK-based version, see the classic documentation.

Transform dense_vector fields to knn_vector

Migration Assistant can automatically convert Elasticsearch dense_vector mappings into OpenSearch knn_vector mappings during metadata migration. The target mapping must be valid for the OpenSearch k-NN model, and the application may need query changes after migration.

Built-in transformation behavior

The metadata migration path can:

• Convert dense_vector to knn_vector.
• Translate vector dimensions and related settings.
• Prepare the target mapping for OpenSearch vector search.

Depending on the target, additional vector compatibility transforms may also apply, including Serverless-specific adjustments.

Identifying dense_vector fields

To verify whether your source uses dense_vector fields, run the following command:

console clusters curl source /_mapping

If the source mapping contains "type":"dense_vector", inspect those indexes carefully during assessment and pilot validation.

Post-migration validation

Verify the target mapping:

console clusters curl target /your-index/_mapping
workflow show

Additionally, validate that the target cluster supports the vector search features you intend to use.

Application impact

Even when the mapping migration succeeds, query behavior may still need to change. Validate the search layer carefully if your application currently relies on Elasticsearch vector query patterns.

Additional considerations

Run a pilot migration with representative queries before cutover in the following cases:

• The target is OpenSearch Serverless.
• The target version has vector-engine compatibility constraints.
• The application relies on specific vector query syntax or ranking behavior.