Skip to Content
We are live but in Staging 🎉
Quickstart

Quickstart

From nothing to a working vector search in a few minutes. You’ll allocate a database (control plane), then use the Milvus SDK for everything else (data plane).

Before you start

  • A Dodil IAM service account (a client_id + client_secret) and an access token from it — get one with the client-credentials flow  (or dodil login). That same token authenticates both the control plane and the Milvus connection; see Auth and Access.
  • The dodil CLI (for the quickest path) — see Install the CLI . Or call the control-plane API directly.
  • pip install pymilvus for the SDK steps.

1. Allocate a database

dodil vbase db create my-vectors

Allocation is asynchronous — the database starts in CREATING and becomes RUNNING within moments. (Over the API: AllocateDatabase then poll GetService until RUNNING.)

2. Get its connection details

dodil vbase db use <service-id>

This resolves the database’s endpoint, port, and db_name (via GetServiceAccess) and stores them so the CLI — and you — can connect.

3. Connect the Milvus SDK

from pymilvus import MilvusClient
 
client = MilvusClient(
    uri="https://<endpoint>:443",   # endpoint + port from step 2
    token="<IAM access token>",      # your IAM token is your Milvus token
    db_name="<db_name>",             # the allocated database
)

See Connecting with the Milvus SDK for the full contract (TLS, tokens, other languages).

# create a collection with a 768-dim vector
client.create_collection(collection_name="docs", dimension=768)
 
# insert a few vectors
client.insert(collection_name="docs", data=[
    {"id": 1, "vector": [0.1] * 768, "text": "hello"},
    {"id": 2, "vector": [0.2] * 768, "text": "world"},
])
 
# search
hits = client.search(
    collection_name="docs",
    data=[[0.15] * 768],
    limit=3,
    output_fields=["text"],
)
print(hits)

That’s standard Milvus 2.6 — collections, indexes, and search are all the Milvus SDK. For schemas, custom indexes (HNSW/IVF), hybrid + BM25 search, and more, see the Recipes and the Milvus docs .

Next

OverviewCore Concepts