Open Source Summit India 2026

Every time one Pod talks to another in Kubernetes, the Linux kernel does a surprising amount of work.

Engineers know Pods get IP addresses. They know Services like ClusterIP and NodePort make workloads reachable. And they trust that traffic somehow finds the right destination.

But what actually happens to a packet once it leaves a Pod, especially when it needs to reach another Pod on a different node?

In this session, we trace that journey across a live Kubernetes cluster. We follow real Pod-to-Pod traffic, observe how packets move across nodes using native Linux networking primitives, and examine what enables flat, routable Pod networking without NAT between workloads.

Rather than treating Kubernetes networking as magic, we connect what we see to the underlying constructs like network namespaces, veth pairs, bridges, routing tables, and packet filtering rules.

Through live demos, attendees will build a clear, practical mental model of how Kubernetes moves packets and leave with a clear mental model for explaining, observing, and debugging Pod-to-Pod traffic in Kubernetes.