Taxonomy of Docker

This course covers the core concepts behind containerizing an app: taxonomy, writing a Dockerfile, volumes, Compose, and secrets. All exercises run in the browser — no local Docker install required.

Docker has four core building blocks — the Dockerfile, the image, the container, and the daemon — and they're easy to confuse. The rest of this course builds on telling them apart, so it's worth getting them straight first.

The Docker daemon

When you type a docker command, you're talking to the Docker daemon (dockerd), a background process that does the actual work: building images, running containers, pulling from registries Registry An online store of prebuilt images. When an image is not already on your machine, Docker downloads it from here. , and managing volumes. The docker CLI is just a thin client that sends it instructions. It runs in the background on your laptop and on the server in production. Knowing it exists explains why Docker commands work from any directory, and why restarting Docker Desktop fixes mysterious build errors.

The host machine

The host (or host machine) is the computer Docker runs on: your laptop in development, a server in production. The daemon, your images, and your containers all live on it. A container is isolated from the host by default, with its own filesystem and network, but the two can be connected on purpose: publishing a port (-p 3000:3000) links a host port to a container port, and mounting a volume links a host folder to a container path. When a lesson says "the host machine," it means this underlying computer, as opposed to the isolated environment inside a container.

Image vs. Container vs. Dockerfile

• Dockerfile: a text file of instructions for building an image.
• Image: a packaged, runnable snapshot of an application and everything it needs to run — code, runtime, config. It could be something you built from a Dockerfile, or a pre-built image pulled from a registry like Docker Hub Docker Hub The default public registry. Official images like node and postgres live here, ready to pull. (a database, a web server, etc.).
• Container: a running instance of an image. Multiple containers can run at the same time — from the same image or from different images — each isolated from the others.

The part that trips people up: a Dockerfile builds an image, and an image runs as a container. Editing the Dockerfile changes nothing until you rebuild, and a rebuild doesn't touch a running container until you start a new one.