Connection Endpoints

Zilliz Cloud exposes three endpoints, each with distinct responsibilities.

	Control Plane API Endpoint	Project Endpoint (On-Demand)	Real-time Serving Endpoint
URL Pattern	https:*//*api.cloud.zilliz.com	https:*//*{project-id}.{region}.api.zillizcloud.com	https:*//*{cluster-id}.{region}.vectordb.zillizcloud.com:19530
Responsibility	Resource lifecycle: clusters, volumes, jobs, and all other control plane activities	Data import, batch search	Full Collection API (DDL + DML + DQL)
Data Operations	None (except data import)	Bulk-insert and import; search billed by CU	Insert, upsert, and delete with low-latency search and query
When to use	Provisioning infrastructure and automation	Batch processing, exploration, validation, experiments	Production serving, always-on low-latency queries

Connect to a real-time serving cluster

Zilliz Cloud offers the following types of serving clusters: Free, Serverless, and Dedicated. You need to follow the examples below to set up connections.

from pymilvus import MilvusClient

# connect to a dedicated cluster
client = MilvusClient(
    uri="https://{cluster-id}.{region}.vectordb.zillizcloud.com:19530",
    token="YOUR_API_KEY"
)

# connect to a free / serverless cluster
client = MilvusClient(
    uri="https://{cluster-id}.serverless.{region}.vectordb.zillizcloud.com",
    token="YOUR_API_KEY"
)

You can use a valid API key with appropriate permissions or a cluster credential in username:password format as the authentication token.

Connect to an on-demand cluster

Zilliz Cloud provides the session object, which you can use to attach an on-demand cluster to a database and conduct searches in the database.

from pymilvus import MilvusClient

client = MilvusClient(
    uri="https://{project-id}.{region}.api.zillizcloud.com",
    cluster="inxx-xxxxxxxxxxxxxxx",
    token="YOUR_API_KEY"
)

session = client.session(cluster_id="inxx-xxxxxxxxxxxxxx")

# Then, use session to conduct DQL operations, such as query, get, search, and hybrid_search.

When connecting to an on-demand compute endpoint, you also need to set the cluster ID of an on-demand cluster so that you can use the compute resources in that cluster to perform searches and queries.

You should use a valid API key with sufficient permissions as the authentication token when you connect to a project endpoint.

Connect to Zilliz Cloud Control Plane API endpoint

When you need to create clusters and volumes, or manage control-plane resources such as backups, restores, and migrations, use the platform endpoint.

For example, you can list available cloud providers as follows:

export BASE_URL="https://api.cloud.zilliz.com"
export TOKEN="YOUR_API_KEY"

curl --request GET \
--url "${BASE_URL}/v2/clouds" \
--header "Authorization: Bearer ${TOKEN}" \
--header "Content-Type: application/json"

For details, refer to RESTful API Reference.