Postgres

Configure your own Postgres instance as data sink.

Configure via UI

Configure Parameters

To configure Postgres as a vector store for your LlamaCloud documents you will need the following:

Parameter	Description	Example
Database	Database name	llamaindex
Host	Connection endpoint	my-postgres-cluster.us-east-1.rds.amazonaws.com
User	Database username	postgres
Password	Password for database user	*****
Table Name	Table where embeddings will be stored	llamaindex
Schema Name	Schema in which the database table will exist	public
Embedding Dimension	Dimension size of embeddings	1536
Port	Port where Postgres listens	5432

Configure via API / Client

Python Client

from llama_cloud.types import CloudPostgresVectorStore

ds = {
    'name': '<your-data-sink-name>',
    'sink_type': 'POSTGRES',
    'component': CloudPostgresVectorStore(
        database='<database-name>',
        host='<database-host>',
        user='<user>',
        password='<password>',
        port=5432,
        embed_dim=1536,
        schema_name='<schema>',
        table_name='<table>'
    )
}

data_sink = client.data_sinks.create_data_sink(request=ds)

Typescript Client

const ds = {
    'name': '<your-data-sink-name>',
    'sinkType': 'POSTGRES',
    'component': {
        'database': '<database-name>',
        'host': '<database-host>',
        'user': '<user>',
        'password': '<password>',
        'port': 5432,
        'embed_dim': 1536,
        'schema_name': '<schema>',
        'table_name': '<table>'
    }
}

data_sink = await client.dataSinks.createDataSink({
projectId: projectId,
body: ds
})