Docker & Containers

Containers solved a specific, miserable problem. Software depends on the machine it runs on — a library version, an environment variable, a directory that happens to exist — and none of that travels with the code. So it works for the person who wrote it and fails for everyone else, and the resulting afternoon of debugging teaches nobody anything.

A container packages the application together with the parts of the system it needs, so the thing that ran on a laptop is the same thing that runs on the server. That is the entire idea. Everything in this course is a consequence of it: images and layers are how the package is built, Compose is how several packages talk to each other, and Kubernetes is what happens when there are hundreds of them and nobody can start them by hand.

The course is built around Nana's three-hour tutorial, which is unusually good at explaining why each piece exists before showing the commands. Mosh's course covers the same ground with different emphasis and a lot more attention to writing a good Dockerfile, which is where most real Docker time gets spent.

Kubernetes is included but kept in its place. Most people do not need it and adopting it early is a well-documented way to make a simple system complicated. The hundred-second lesson is required so you know what it is; the four-hour course is there for when a real problem asks for it. About four hours required, and around thirty-seven in the pool — three more full courses, a module on writing images properly, and a CI/CD pipeline at the end. Install Docker and run every command — containers are cheap to break and free to recreate, which makes them unusually pleasant to learn by experiment.