Skip to main content

Kubernetes API Objects

Updated Apr 07, 2022 ·

Pods​

A Pod is the basic deployment unit in Kubernetes, representing one or multiple containers that run together.

  • Atomic scheduling unit
  • Represents container-based applications
  • Defines required resources

Important concepts:

  • Ephemeral: Pods are not reused; a new pod is created if needed
  • Atomicity: If a container fails, the entire pod is terminated

Kubernetes manages the pod by monitoring:

  • State – Is the pod running?
  • Health – Is the app inside the pod active?
  • Liveness probes – Are expected responses returned?

Each Pod has a single IP address:

  • Containers in the same Pod share the IP
  • Containers communicate via localhost
  • Pods coordinate port assignments

Controllers​

Controllers maintain the desired system state.

  • Creates and manages Pods
  • Ensures desired state is achieved
  • Monitors and adjusts for pod health

Common Controllers:

  • ReplicaSet – Keeps a specified number of pod replicas running
  • Deployment – Manages changes between ReplicaSets

Services​

Services offer stable access points for applications provided by Pods, adding persistence to the system.

  • Networking layer for Pod access
  • Allocates an IP and DNS for the service
  • Automatically updates routing for redeployed Pods
  • Scales apps by adjusting Pods
  • Enables load balancing across Pods

A Service uses a virtual IP address:

  • Maps to multiple Pods
  • Allows external services to use a single IP

Storage​

Storage objects provide persistent storage for data.

  • Volumes – Tightly coupled storage with physical media for Pods
  • Persistent Volumes – Cluster-level storage, independent of Pods, using a Persistent Volume Claim