Two teams create AKS clusters on the same day. One discovers that its on-premises services can address pods directly. The other sees only node or load-balancer addresses. Both clusters are working ...

Introduction In Create a Kubernetes CNI Plugin from Scratch, we built a bridge CNI in Bash or Go. It creates a veth pair for each Pod, delegates address allocation to host-local, and connects the ...

Introduction A Kubernetes Pod starts inside a new network namespace with no useful network interface. The container runtime then executes a Container Network Interface (CNI) plugin and asks it to ...

Introduction When an application connects to a Kubernetes Service, the first useful interception point can occur before a packet exists. Cilium’s socket load balancer attaches eBPF programs to cgr...

Introduction Welcome to Part 5, the final chapter of our Kubernetes Networking series. We have journeyed through the entire stack: Part 1: The Model (IP-per-Pod). Part 2: The CNI (Plumbing)...

Introduction hotspot-bpf uses eBPF to turn raw kernel events into real-time performance explanations. It correlates CPU time, scheduler contention, and page-fault pressure in a single window, reve...

Introduction Welcome to Part 4 of our Kubernetes Networking series. So far, we have built a solid foundation: Part 1: Every Pod gets a unique IP. Part 2: CNI plugins connect these Pods. P...

Introduction Welcome to Part 3 of our Kubernetes Networking series. In Part 1, we defined the Model: every Pod gets an IP. In Part 2, we explored the CNI: how those IPs are assigned and how packe...

Introduction In Part 1, we established the Model: the “Golden Rules” of Kubernetes networking. Before exploring how connections are made, let’s briefly revisit these fundamental design principles....

Introduction Kubernetes networking is often described as confusing or even “black magic.” For many people, that confusion starts the moment something simple doesn’t work. Example 1: Pods that can...