Container Orchestration (Kubernetes basics)

Kubernetes (K8s) is a container orchestration platform that automates the deployment, scaling, networking, and lifecycle management of containerized applications across clusters of machines. It abstracts away individual servers and presents a cluster as a single computational surface, using declarative configuration (YAML manifests) to define the desired state of applications. The Kubernetes control plane continuously reconciles actual state with desired state.

The core abstractions include Pods (the smallest deployable unit, typically one container), Deployments (managing ReplicaSets for rolling updates and rollbacks), Services (stable network endpoints with load balancing across Pods), ConfigMaps and Secrets (externalized configuration), and Ingress resources (HTTP routing and TLS termination). Namespaces provide logical isolation within a cluster, and resource quotas enforce CPU and memory limits per namespace.

For most teams, managed Kubernetes services (EKS, GKE, AKS) eliminate the complexity of running the control plane. Key operational concerns include defining proper resource requests and limits, configuring liveness and readiness probes, setting Pod Disruption Budgets for safe node maintenance, and using Horizontal Pod Autoscalers to scale based on CPU, memory, or custom metrics. Helm charts or Kustomize overlays are commonly used to template and manage Kubernetes manifests across environments.