Buckets are the entry point for ingestion. They define a schema for the objects you register and provide a logical container for organizing raw files (blobs) before processing.
Overview
- Purpose: Store registered objects (and their blobs) under a clear schema.
- Schema: Enforces allowed field names and types for blobs.
- Isolation: Use
X-Namespace
to scope buckets per tenant/environment. - Processing: Bucket creation and object registration do not process data; collections handle processing later.
Bucket schema
Define allowed fields and types. Field types align with ingestion inputs (e.g.,image
, video
, pdf
, document
, json
).
bucket_schema.properties
: keys are blob field names; values specify the type.- Objects referencing missing or mismatched fields are rejected.
Create a bucket
- API: Create Bucket
- Method: POST
- Path:
/v1/buckets
- Reference: API Reference
Manage buckets
Behavior & validation
- Schema enforcement: Objects must conform to
bucket_schema
. - Naming:
bucket_name
must be unique within a namespace. - Schema mutability: Schema is set at creation; updates are limited to mutable fields (name, description, metadata).
- Rate limits: Bucket operations are rate-limited per account tier.