feed-operations
Installation
SKILL.md
Vespa Feed Operations
Overview
Vespa's feeding system is the pipeline through which documents enter, get updated in, and are removed from a content cluster. Every write travels through a document processing chain before being distributed to the appropriate content nodes via the distributor. The feeding layer supports individual CRUD operations as well as high-throughput bulk ingestion.
Key concepts:
- Document: a typed data record identified by a globally unique document ID.
- Put: insert or fully replace a document.
- Update: partially modify fields of an existing document.
- Remove: delete a document.
- Feed client: any component that submits document operations — the CLI, Java client, or raw HTTP calls.
Document ID Format
Every document in Vespa is addressed by a document ID string with the following structure: