Skip to main content

Embeded Milvus

Project description

Milvus Lite

PyPI Version


Milvus Lite is a lightweight version of Milvus that can be embedded into your Python application. It is a single binary that can be easily installed and run on your machine.

It could be imported as a Python library, as well as use it as a command standalone server.

Thanks to Milvus standalone version could be run with embedded etcd and local storage, Milvus Lite does not have any other dependencies.

Everything you do with Milvus Lite, every piece of code you write for Milvus Lite can be safely migrated to other forms of Milvus (standalone version, cluster version, cloud version, etc.).

Please note that it is not suggested to use Milvus Lite in a production environment. Consider using Milvus clustered or the fully managed Milvus on Cloud.


Milvus Lite is available in:

  • Google Colab example
  • Jupyter Notebook

Here's also a list of verified OS types where Milvus Lite can successfully build and run:

  • Ubuntu >= 18.04 (x86_64)
  • CentOS >= 7.0 (x86_64)
  • MacOS >= 11.0 (Apple Silicon)


  • For linux we use manylinux2014 as the base image, so it should be able to run on most linux distributions.
  • Milvus Lite can also run on Windows. However, this is not strictly verified.


Milvus Lite is available on PyPI. You can install it via pip for Python 3.6+:

$ python3 -m pip install milvus

Or, install with client(pymilvus):

$ python3 -m pip install "milvus[client]"

Note: pymilvus now requires Python 3.7+


Import as Python library

Simply import milvus.default_server.

from milvus import default_server
from pymilvus import connections, utility

# (OPTIONAL) Set if you want store all related data to specific location
# Default location:
#   %APPDATA%/milvus-io/milvus-server on windows
#   ~/.milvus-io/milvus-server on linux
# default_server.set_base_dir('milvus_data')

# (OPTIONAL) if you want cleanup previous data
# default_server.cleanup()

# Start you milvus server

# Now you could connect with localhost and the given port
# Port is defined by default_server.listen_port
connections.connect(host='', port=default_server.listen_port)

# Check if the server is ready.

CLI milvus-server

You could also use the milvus-server command to start the server.

$ milvus-server

The full options cloud be found by milvus-server --help.

Advanced usage

Debug startup

You could use debug_server instead of default_server for checking startup failures.

from milvus import debug_server

and you could also try create server instance by your self

from milvus import MilvusServer

server = MilvusServer(debug=True)

If you're using CLI milvus-server, you could use --debug to enable debug mode.

$ milvus-server --debug

Configurations for Milvus

Milvus Lite could set configure by API as well as by CLI. We seperate the configurations into two parts: basic and extra.

The basic configurations

You could find available configurations by milvus-server --help for got the list of basic configurations.

These basic configurations including:

  • Some listen ports for service, e.g. --proxy-port for specifying the port of proxy service.
  • Some storage configurations, e.g. --data for specifying the data directory.
  • Some log configurations. e.g. --system-log-level for specifying the log level.

If you using Python API, you could set these configurations by MilvusServer.config.set method.

# this have the same effect as `milvus-server --system-log-level info`
default_server.config.set('system_log_level', 'info')

All configuable basic configurations could be found in config yaml template, which is installed with milvus package.

The extra configurations

Other configurations are extra configurations, which could also be set by MilvusServer.config.set method.

for example, if we want to set dataCoord.segment.maxSize to 1024, we could do:

default_server.config.set('dataCoord.segment.maxSize', 1024)

or by CLI:

milvus-server --extra-config dataCoord.segment.maxSize=1024

Both of them will update the content of Milvus config yaml with:

    maxSize: 1024


You could close server while you not need it anymore. Or, you're able to using with context to start/stop it.

from milvus import default_server

with default_server:
    # milvus started, using default server here

Data and Log Persistence

By default all data and logs are stored in the following locations: ~/ (VERSION is the versiong string of Milvus Lite).

You could also set it at runtime(before the server started), by Python code:

from milvus import default_server

Or with CLI:

$ milvus-server --data milvus_data

Working with PyMilvus

Milvus Lite could be run without pymilvus if you just want run as a server. You could also install with extras client to get pymilvus.

$ python3 -m pip install "milvus[client]"


Milvus Lite is friendly with jupyter notebook, you could find more examples under examples folder.


If you want to contribute to Milvus Lite, please read the Contributing Guide first.

Report a bug

When you use or develop milvus-lite, if you find any bug, please report it to us. You could submit an issue in milvus-lite or report you milvus repo if you think is a Milvus issue.


Milvus Lite is under the Apache 2.0 license. See the LICENSE file for details.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distribution

milvus-2.3.8-py3-none-macosx_12_0_arm64.whl (36.3 MB view hashes)

Uploaded Python 3 macOS 12.0+ ARM64

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page