Streaming API

Cosdata’s streaming data ingestion is designed for scenarios where individual records must become immediately searchable upon insertion. This is ideal for real-time monitoring systems, live content feeds, and streaming analytics.

Streaming ingestion uses a simplified fire-and-forget API that abstracts away transactional complexity. Each operation is atomic at the record level and immediately available for querying.

Streaming Data: Upsert Vectors

This API provides a simplified fire-and-forget way to upsert vectors, abstracting away the complexity of atomic operations. The entire workflow is handled in a single request.

Endpoint: POST /vectordb/collections/{collection_id}/streaming/upsert

URL Parameters:

Parameter	Type	Required	Description
collection_id	string	Yes	ID (name) of the collection.

Request Body:

{
  "vectors": [
    {
      "id": "vector_id_1",
      "dense_values": [0.1, 0.2, 0.3, ...]
    },
    {
      "id": "vector_id_2",
      "text": "Document text content"
    }
  ]
}

Status Codes:

Code	Description
200	Success. Vectors upserted successfully.
404	Not Found. Collection not found.
500	Server Error. Failed to upsert vectors.

---

Streaming Data: Delete a Vector

This API provides a simplified fire-and-forget way to delete a vector, abstracting away the complexity of atomic operations. The entire workflow is handled in a single request.

Endpoint: DELETE /vectordb/collections/{collection_id}/streaming/delete/{vector_id}

URL Parameters:

Parameter	Type	Required	Description
collection_id	string	Yes	ID (name) of the collection.
vector_id	string	Yes	ID of the vector to delete.

Status Codes:

Code	Description
204	No Content. Vector deleted successfully.
404	Not Found. Collection or vector not found.
500	Server Error. Failed to delete vector.