Kubernetes Overview and Key Concepts for Deploying Java Services

Overview

In Java development, the docker run command together with a private Docker registry can quickly deploy a Java service, but it has several drawbacks: a single container runs on one machine, offering no high availability, and services often depend on other components like databases or message queues that need to be deployed together.

Docker Compose helps during development by allowing a set of containers to be defined in a YAML file and started or stopped as a group, yet it is still limited to a single host.

For reliable production deployment, a Docker orchestration framework such as Kubernetes is required.

Kubernetes Introduction

Kubernetes is a Docker orchestration platform that abstracts a pool of hardware resources into a single resource pool for running services, maintaining the desired number of service instances and ensuring they stay online even if containers or machines fail.

Kubernetes provides three main functions:

Resource management – treats a cluster of machines as a single pool of CPU, memory, and storage.

Scheduling – selects the nodes on which containers run, considering resource needs and affinity/anti‑affinity rules.

Service management – offers naming, versioning, load balancing, rolling updates, and rollback capabilities for services.

Kubernetes Architecture

Kubernetes runs on a set of machines divided into master nodes and worker (ordinary) nodes. Typically there are a few master nodes and many worker nodes.

Master node manages the cluster, while worker nodes (also called "nodes") run one or more Pods.

Master node components include:

API server : REST API for deploying and managing services, used by kubectl.

etcd : NoSQL key‑value store for cluster data.

Scheduler : Chooses nodes for Pods.

Controller manager : Runs controllers (e.g., replication controller) to keep the actual state matching the desired state.

Worker node components include:

Kubelet : Creates and manages Pods on the node.

Kube‑proxy : Handles networking and load balancing across Pods.

Pods : The actual application services.

Key Kubernetes Concepts

Understanding a few core objects makes using Kubernetes efficient. The most important objects are:

Pod : The basic deployment unit, consisting of one or more containers sharing an IP address and storage. Pods may include sidecar containers for auxiliary tasks. Pods are short‑lived because they can be terminated if the container or node fails.

Deployment : Declarative specification for Pods. A Deployment controller ensures the desired number of Pod replicas are running, supports rolling updates, and can roll back to previous versions.

Service : Provides a stable network endpoint (IP and DNS name) for a set of Pods, enabling load balancing for TCP/UDP traffic. By default it uses ClusterIP; external access requires NodePort or LoadBalancer.

ConfigMap : Stores configuration key‑value pairs that can be injected into Pods as environment variables or configuration files. Sensitive data should be stored in Secrets, which are a specialized form of ConfigMap.