Understanding Containers and Container Management Platforms: From Concepts to Hands‑On Docker/Kubernetes Deployment

The development of cloud computing follows the pattern “division leads to union, union leads to division,” and containers emerged to solve the rapid, batch deployment challenges of the PaaS layer that cloud elasticity alone could not address.

Using the shipping‑container analogy, the article shows how a simple Java web site with many scattered dependencies can be packaged into an isolated environment; isolation is achieved through Linux namespace (separate network, PID, user space) and cgroup (resource limits).

Containers are built from immutable images that capture the filesystem state at a point in time; unlike heavyweight virtual‑machine images, container images are small because they share the host kernel. Docker, AppC, and Mesos Container are examples of runtimes, and the term “container” is not synonymous with Docker.

Three major container orchestration platforms are introduced: Kubernetes (derived from Google’s Borg), Mesos (used by Twitter), and Docker Swarm. Their design philosophies, community activity, and scalability characteristics are compared.

The article also highlights advanced platform features such as self‑discovery (service naming), self‑healing (automatic restart on failure), and auto‑scaling (dynamic resource adjustment).

For Docker beginners, the lifecycle is explained: images ↔ containers, with commands like docker run, docker stop, docker start, docker commit, and building images via a Dockerfile. A minimal Dockerfile example is shown:

FROM tomcat:8.0-jre7
ADD springblog.war /usr/local/tomcat/webapps/
ADD spring-instrument-tomcat-4.2.2.RELEASE.jar /usr/local/tomcat/lib/
ENTRYPOINT ["catalina.sh", "run"]

The practical case walks through uploading a WAR and JAR to object storage, writing the Dockerfile, building the image, creating a stateless Tomcat service, configuring a load balancer, and finally accessing the deployed blog application.