Images, layers, and registries — Docker foundations

An image is a read-only stack of filesystem layers plus metadata for how to start a container. Each Dockerfile instruction usually creates a new layer, and Docker reuses unchanged layers to avoid rebuilding everything.

Registries store and distribute images. Tags are convenient names like latest or 1.2.0, but they can be moved; digests identify exact image content and are safer when repeatability matters.

Image layers

A Docker image is built as a sequence of filesystem changes. Installing packages, copying files, and setting permissions become part of the image history.

When a layer has not changed, Docker can reuse it. This is the reason Dockerfile order directly affects build speed.

Tags and digests

A tag is a human-friendly pointer. A digest is a content address. Use tags while learning and use pinned versions or digests when reproducibility matters.

docker pull ubuntu:24.04
docker image ls ubuntu
docker image inspect ubuntu:24.04
docker history ubuntu:24.04

Registry workflow

A registry is where images are stored. Pulling downloads missing layers; pushing uploads layers the registry does not already have.

In teams, the registry becomes the handoff point between build, CI, deployment, and rollback.

what to remember

Pin base images deliberately when the same build must behave the same way later.
Small layers are useful, but predictable dependency order matters more than clever one-liners.
When a pull is slow, Docker is usually downloading only the layers it does not already have.