-
Notifications
You must be signed in to change notification settings - Fork 2.5k
Description
Is your feature request related to a problem? Please describe
Currently, document fields are ingested from the request source payload via the XContent parsing pipeline before being consumed by field mappers.
For high-performance ingestion scenarios (e.g., protobuf/gRPC), we need the ability to pass additional typed field values alongside the document source during indexing.
This issue proposes extending IndexRequest / SourceToParse to carry optional ExtraFieldValues so selected mappers can consume them directly, bypassing JSON parsing for those fields and reducing CPU/allocation overhead. This improves indexing throughput for large structured fields such as dense vectors.
Describe the solution you'd like
Additional field values would come from ExtraFieldValues
Related component
Indexing
Describe alternatives you've considered
No response
Additional context
[Feature Request][gRPC] Move primitive array types out of source](#19638)