Skip to main content

Kubernetes Beginnings

What is Kubernetes?​

You have containerized applications but you need to manage them! Kubernetes can do that for you.

Overview​

  • Kubernetes is an open-source tool for automating deployment, scaling, and operation of containerized applications
  • Originated from Google's Borg system
  • Designed for large-scale management with faster releases and self-healing capabilities
  • Operates as a distributed system, forming clusters from physical or virtual machines
  • Can run on-premises or in the cloud
  • Uses scheduling algorithms for container allocation and adapts as needed
  • Has modular design that supports multiple container runtimes, not only Docker

Key Features​

  • Declarative configuration - Define desired state
  • Horizontal scaling - Scale applications
  • Secret management - Handle sensitive data
  • Service discovery and load balancing - Connect services automatically
  • Linux and Windows container support - Multi-platform
  • Stateful application support - Handle databases and persistent apps
  • Persistent volumes - Storage management
  • Log collection - Centralized logging
  • CPU/Memory quotas - Resource management
  • Batch processing - Run jobs and cron jobs
  • RBAC - Role-based Access Control

Alternatives to Kubernetes​

  • DC/OS
  • Docker Swarm
  • Amazon ECS

How to Deploy Kubernetes​

Deployment Methods​

Single-node clusters (for development and testing):

  • Docker for Mac/Windows
  • minikube
  • kubeadm

Kubernetes in Docker (kind):

  • Recommended for creating ephemeral clusters in CI pipelines

Multi-node clusters:

  • Advised for production to benefit from horizontal scaling and fault tolerance

Single-node vs Multi-node​

  • Single-node clusters - Simpler, suitable for learning, development, and testing
  • Multi-node clusters - Offer resilience and scalability for production workloads

Managed Kubernetes Services​

Cloud Services:

  • Amazon EKS, Azure AKS, Google GKE
  • Simplify Kubernetes deployment but may have delayed updates

Self-managed Tools:

  • kubespray, kops, kubeadm
  • Offer more customization and control

Deployment Considerations​

  • Level of control needed
  • Cloud provider investments
  • Need for enterprise support
  • Concerns about vendor lock-in

Flexibility​

Kubernetes supports hybrid cloud environments and both Linux and Windows containers, offering deployment flexibility.

Nodes​

What are Nodes?​

  • Virtual identities assigned by Kubernetes to systems in the cluster
  • Can be Virtual Machines, bare-metal servers, containers, etc.
  • Each node has unique identity for resource accounting and monitoring
  • Used for workload management throughout cluster

Node Components​

  • kubelet - Node agent that manages local workloads
  • kube-proxy - Node agent that manages network traffic
  • Container runtime - Required to run containerized workloads

Node Agent Responsibilities​

  • Interact with runtime to run containers
  • Monitor containers and node health
  • Report issues and node state to API Server
  • Manage network traffic to containers

Node Types​

Control Plane Nodes:

  • Run control plane agents (API Server, Scheduler, Controller Managers, etcd)
  • Also run kubelet, kube-proxy, container runtime
  • Include add-ons for networking, monitoring, logging, DNS
  • At least one required, multiple for High Availability

Worker Nodes:

  • Run kubelet, kube-proxy, container runtime
  • Include add-ons for networking, monitoring, logging, DNS
  • Provide resource redundancy in cluster
  • One or more typically included

Hybrid/Mixed Nodes:

  • Single all-in-one cluster on single system
  • Host both control plane agents and user workloads
  • Used when HA and resource redundancy not important
  • Good for learning and development

Node Identity Creation​

  • Created during cluster bootstrapping process
  • Minikube uses kubeadm bootstrapping tool
  • Init phase - Initialize control plane node
  • Join phase - Add worker or control plane nodes

Cluster Distribution​

  • Nodes can be on same private network
  • Can span across different networks
  • Can span across different cloud networks
  • Control plane + worker nodes = Kubernetes cluster

Kubernetes Architecture​

For detailed information about Kubernetes architecture, see Kubernetes Architecture.

Quick Architecture Overview​

  • Control Plane - Manages the cluster (API Server, Scheduler, Controller Manager, etcd)
  • Worker Nodes - Run your applications (kubelet, kube-proxy, container runtime)
  • Pods - Smallest deployable units that contain containers
  • Services - Provide stable network endpoints for Pods
  • Deployments - Manage Pod replicas and updates

Installing Kubernetes​

Using Kubeadm on EC2​

Installing a Kubernetes cluster using Kubeadm on EC2 provides insight into how control plane and worker node components work internally, including the join phase mechanism to connect master and worker nodes.

Using Minikube​

Minikube is great for local development. Behind the scenes, Minikube works like Kubeadm and performs the join phase mechanism.

Minikube Commands​

# Bootstrap a single node cluster with LTS Kubernetes
minikube start

# Specify a specific Kubernetes version
minikube start --kubernetes-version=v1.31.1

# Use a specific driver (if you have virtualization drivers)
minikube start --driver=virtualbox

# Check the status
minikube status

# Stop the minikube cluster
minikube stop

# Delete minikube and its VM completely
minikube delete

Additional Minikube Commands​

# Get minikube IP
minikube ip

# SSH into minikube
minikube ssh

# Open Kubernetes dashboard
minikube dashboard

# List available addons
minikube addons list

# Enable an addon (e.g., ingress)
minikube addons enable ingress

# View minikube logs
minikube logs

Accessing Kubernetes​

Access Methods​

A Kubernetes cluster can be accessed via:

  • kubectl - CLI tool (most common)
  • Kubernetes Dashboard - Web-based GUI
  • APIs - Direct API calls for programming or CLI

Basic kubectl Commands​

# Check cluster info
kubectl cluster-info

# Get cluster nodes
kubectl get nodes

# Get all resources
kubectl get all

# Check kubectl version
kubectl version

# Get cluster configuration
kubectl config view