2learn

Kubernetes

k8s

What it is

Kubernetes is an orchestrator: you declare how many copies of a container should run, how they should be reachable and what resources they need, and it works continuously to make reality match. It schedules containers across a pool of machines, restarts what fails, replaces what disappears and rolls out new versions gradually. It is a platform for running other people’s software, and most of what it does is failure handling.

Why it matters

Running containers by hand on a handful of machines stops working as soon as a machine dies, a version must be rolled out without downtime, or demand changes faster than someone can log in. Kubernetes automates exactly those problems, and it does so with a declarative interface: the desired state lives in version control, and the cluster is a reconciliation loop away from matching it. The cost is substantial — a cluster is a distributed system with its own vocabulary, failure modes and operational burden — so it pays off when the workload is genuinely dynamic or the team needs a uniform deployment target, and it is heaviness when a single container on a single host would do.

In this track

  1. The object model (06-cloud-devops/14-kubernetes/01-the-object-model) — pods, deployments and services, and how they compose.
  2. Scheduling and scaling (06-cloud-devops/14-kubernetes/02-scheduling-and-scaling) — how work is placed on nodes and how capacity grows and shrinks.
  3. Configuration and secrets (06-cloud-devops/14-kubernetes/03-configuration-and-secrets) — separating the image from the environment it runs in.
  4. Self-healing and observability (06-cloud-devops/14-kubernetes/04-self-healing-and-observability) — what the cluster does when things break, and how you find out.

Prerequisites

Go deeper

Analytics cookies are set only if you allow them. Nothing is loaded, and nothing is sent, until you choose.