Docker Compose is an Apache-2.0 licensed Go tool that defines and runs multi-container applications on Docker from a single Compose file, aimed at developers and operators who need several containers to start together as one isolated application.

What it is

Docker Compose is a tool for running multi-container applications on Docker, where the application is described in a Compose file. That file defines how one or more containers that make up the application are configured, and once it exists the whole application can be created and started with a single command, docker compose up. The project is written in Go, lives in the Docker ecosystem, and is published under the Apache-2.0 licence.

The concrete problem it solves is the manual coordination of several containers that belong to one application. Without a Compose file, each container has to be launched, networked and configured separately, and the resulting environment is hard to reproduce. Compose replaces that with a declarative file, normally compose.yaml, which is paired with a Dockerfile so the application environment can be reproduced anywhere and the services run together in an isolated environment. It also replaces the legacy Python implementation of Compose, which remains available only under the v1 branch.

Key capabilities

• Runs a full multi-container application with one command, docker compose up.
• Defines services in the Compose file format maintained at the Compose specification, compose-spec.io.
• Builds services from source through a Dockerfile declared in the Compose file with build: ..
• Maps container ports to the host, for example ports: - "5000:5000".
• Mounts host paths into containers through volumes, for example volumes: - .:/code.
• Pulls prebuilt images for supporting services, for example image: redis.
• Ships on Windows and macOS inside Docker Desktop, and on Linux as a standalone binary renamed to docker-compose and placed in a Docker CLI plugins directory such as $HOME/.docker/cli-plugins, /usr/local/lib/docker/cli-plugins or /usr/lib/docker/cli-plugins.

Who uses it and how

• Developers who define an application in a Dockerfile and a compose.yaml, then bring the entire stack up locally in an isolated environment.
• Docker Desktop users on Windows and macOS, where Compose is included and needs no separate installation step.
• Linux users who install the binary from the release page and copy it into a CLI plugins folder, either under the home directory or system-wide, sometimes after making the downloaded file executable with chmod +x.
• Go developers who consume the project as a library, since documentation is published at pkg.go.dev/github.com/docker/compose/v5.
• Contributors to the tool itself, who follow the contributing documentation and report problems on the issue tracker.

Getting started

On Windows and macOS, Docker Compose is included in Docker Desktop, so no separate install is required. On Linux, download the relevant binary from the GitHub release page, rename it to docker-compose, and copy it into $HOME/.docker/cli-plugins or one of the system-wide Docker CLI plugins directories.

How it compares

Docker Swarm used to rely on the legacy Compose file format but never adopted the Compose specification, so some Compose features are not accessible to Swarm users, and after the Mirantis acquisition Swarm is not maintained by Docker Inc. The legacy Python version of Compose is available only under the v1 branch of this repository, while the current Go implementation is the maintained line.

When to use it — and when not to

Anyone self-hosting must still have a working Docker installation, since Compose drives Docker rather than replacing it; the README does not describe any additional database, storage or mail services that Compose itself requires. Swarm users should not expect the newest Compose syntax to be available to them, and teams already committed to a different orchestrator gain little from adding Compose alongside it. The README excerpt is sparse, so readers looking for a full command reference should rely on the project documentation rather than the repository front page.

services:
  web:
    build: .
    ports:
      - "5000:5000"
    volumes:
      - .:/code
  redis:
    image: redis