Course contentsShow
← Software Engineering
Lesson 1444 of 1,93342. Containerization and OrchestrationFree lesson

Pods: The Atomic Unit of Deployment

Understanding pods as groups of containers that share network and storage, and their lifecycle.

Pods: The Atomic Unit of Deployment

What you'll learn: How Kubernetes organizes containers into pods—the smallest deployable units that share resources and have a lifecycle.

What is a Pod?

In Kubernetes, a pod is the smallest unit you can create or deploy. While Docker deals with individual containers, Kubernetes wraps one or more containers into a pod. Think of a pod as a "household" where containers are roommates sharing the same living space—they share the same network address and can access the same storage volumes.

Why Group Containers?

Most commonly, a pod contains a single container running your application. However, some applications need helper containers (called "sidecars") that work closely together. For example, your main web server container might need a logging agent container to ship logs. These containers need to communicate efficiently and share data, so they live in the same pod.

Shared Resources

Containers in a pod share:

  • Network namespace: They communicate via localhost and share the same IP address
  • Storage volumes: They can read/write to the same mounted volumes
  • IPC (inter-process communication): Efficient local communication

Pod Lifecycle

Pods are ephemeral and disposable. When a pod starts, Kubernetes assigns it to a node and it runs until:

  • Its containers complete successfully
  • A container fails and the restart policy says not to retry
  • The pod is deleted manually or by a controller
  • The node fails

When a pod dies, Kubernetes doesn't "revive" it—it creates a brand new pod with a new IP address. This is why pods are considered "cattle, not pets."

Key Takeaway: Pods are Kubernetes' atomic deployment unit, grouping one or more tightly-coupled containers that share networking and storage, with an ephemeral lifecycle managed by the cluster.