Kubernetes Concepts Explained Visually

A Pod is the smallest deployable unit in Kubernetes. It wraps one or more containers that share networking and storage. A bare pod has no controller watching it — if it crashes or its node goes down, nothing recreates it. That is why you use a Deployment to manage pod lifecycle.

A Service provides a stable internal IP address that routes traffic to a set of pods using label selectors. An Ingress sits in front of multiple services and provides external HTTP/HTTPS routing with path-based and host-based rules, typically through a single load balancer.

HPA monitors pod metrics like CPU utilization. When usage exceeds a target threshold (for example 70%), it increases the replica count. When usage drops, it scales back down. HPA adjusts the number of pods, but it cannot provision new cluster nodes — that requires a node autoscaler like Karpenter.

Karpenter is a Kubernetes node autoscaler. When pods are in Pending state because no existing node has enough resources, Karpenter automatically provisions a new node with the right size. When demand drops, it removes underutilized nodes to reduce cost.

Resource Requests tell the scheduler how much CPU and memory a pod needs to be placed on a node. Limits cap the maximum resources a pod can consume. Without limits, a single misbehaving pod can consume all resources on a node, starving every other pod and causing cascading failures.