Intermediate to Advanced|24 hours|44 lessons

Kubernetes Networking

How networking actually works in Kubernetes, from the pod interface to the load balancer and back. The network model and what it delegates, Services and the dataplane, DNS, ingress, NetworkPolicy, multi-cluster, and a full method for debugging the whole chain. CNI agnostic, built for engineers who need to find the broken hop rather than guess at it.

Text-based, no videos

9 modules, 44 lessons

Lifetime access

What you'll learn

Draw the line between what the Kubernetes network model requires and what it delegates entirely to your CNI

Trace a packet from one pod to another, on the same node and across nodes, and name the mechanism at every hop

Choose between overlay encapsulation and native routing, and know what each demands of the underlying network

Size pod CIDRs before address exhaustion becomes a cluster rebuild

Explain what a Service actually is, why no packet is ever delivered to one, and why a pod cannot see the address it was reached through

Compare iptables, IPVS, and eBPF dataplanes, and know what specifically degrades at scale

Diagnose cluster DNS problems that present as application problems, including the ndots latency tax

Curriculum

9 modules · 44 lessons

The Kubernetes Network Model

What Kubernetes actually specifies about networking, what it delegates to the CNI, and how a packet moves between pods on one node and across nodes.

5 lessons

What Kubernetes Actually Specifies25 minFREE What the CNI Decides30 minFREE Pod Networking on a Single Node25 minFREE Pod Networking Across Nodes35 minFREE IP Address Management30 minFREE

Services

What a Service actually is, how the control plane tracks pods behind it, the dataplanes that implement it, and the settings that decide what the backend sees.

6 lessons

What a Service Actually Is30 min EndpointSlices and How Services Track Pods30 min kube-proxy in iptables Mode40 min kube-proxy in IPVS Mode, and Dataplane Alternatives40 min Service Types30 min externalTrafficPolicy and Source IP Preservation25 min

DNS

Cluster DNS from the resolver in the pod to CoreDNS under load, including the resolution behaviour that quietly costs latency.

5 lessons

Cluster DNS and the Service Discovery Model25 min Resolution Behaviour and the ndots Problem35 min dnsPolicy and Custom Resolution25 min Headless Services and StatefulSet DNS20 min DNS Failure Modes and Scaling35 min

Ingress and North South Traffic

Getting traffic from a browser into a pod: the full path, the Ingress resource and why controllers disagree, Gateway API, TLS termination, and load balancer integration.

5 lessons

Getting Traffic Into the Cluster35 min Ingress and the Ingress Controller35 min Gateway API40 min TLS Termination and Certificate Management35 min Load Balancer Integration30 min

NetworkPolicy

The policy model and its switch to default deny, writing selectors that match what you meant, rolling out default deny without breaking DNS, the limits of the standard, and proving a policy actually blocks.

5 lessons

The NetworkPolicy Model25 min Writing Policies That Work35 min Default Deny and Egress Control35 min What NetworkPolicy Cannot Do30 min Testing and Verifying Policy30 min

Multi-Cluster and Advanced Topologies

Reaching across cluster boundaries, discovering services in another cluster, presenting a predictable source address to the outside world, dual stack, and pods that need more than one interface.

5 lessons

Connecting Clusters35 min Cross-Cluster Service Discovery30 min Egress Control and Predictable Source Addresses30 min Dual Stack and IPv635 min Multiple Interfaces and Specialised Networking20 min

Service Mesh and the L7 Layer

Orienting in the mesh decision: what it solves, what it costs, sidecar against sidecarless, workload identity, and where traffic management belongs.

4 lessons

What a Mesh Adds and What It Costs30 min Sidecar and Sidecarless Architectures25 min mTLS and Identity Between Workloads35 min Traffic Management30 min

Debugging Kubernetes Networking

Debugging as a method rather than a list of commands. Walking the chain, reading the symptom, and isolating the broken hop from inside the pod, on the node, and in the packets themselves.

6 lessons

A Systematic Method35 min Which Symptom Points at Which Hop40 min Debugging From Inside the Pod35 min Debugging on the Node40 min Service and DNS Specific Debugging40 min Capture and Observability35 min

Performance and Capstone

Where network overhead comes from, what degrades as the cluster grows, and a full multi-tenant platform network design combining every module.

3 lessons

Network Performance35 min Scaling the Network Layer40 min Capstone: Design the Network for a Multi-Tenant Platform60 min

About the Author

Sharon Sahadevan

AI Infrastructure Engineer

Building production GPU clusters on Kubernetes. H100s, large-scale model serving, and end-to-end ML infrastructure across Azure and AWS.

10+ years designing cloud-native platforms with deep expertise in Kubernetes orchestration, GitOps (Argo CD), Terraform, and MLOps pipelines for LLM deployment.

Author of KubeNatives, a weekly newsletter read by 3,000+ DevOps and ML engineers for production insights on K8s internals, GPU scheduling, and model-serving patterns.

Ready to master this topic?

Start with the free preview lesson and see for yourself.