Foundations Lab  · On-demand

Cilium & Tetragon on Kubernetes (Isovalent) Runtime Protection

Solution overview

Cilium & Tetragon on Kubernetes Runtime Protection

 

The Cilium & Tetragon (Isovalent) Runtime Protection Lab gives you hands-on experience with eBPF-powered networking, observability, and runtime security for Kubernetes. You will work from a web terminal on a dedicated Linux virtual machine that runs **Kubernetes in Docker (KinD)**, so everything stays self-contained and safe to experiment with.

 

 What is eBPF?

 

eBPF (formerly Extended Berkeley Packet Filter) is a technology in the Linux kernel that lets you run small, sandboxed programs directly inside the kernel without modifying kernel source code or loading kernel modules. Programs are verified for safety before loading, then JIT-compiled and attached to hooks such as network events, system calls or function entry points.

 

eBPF programs can observe and act on system activity with high performance and fine-grained visibility while maintaining minimal overhead. eBPF is well suited to networking, security enforcement (why you're here) and observability.

 

Why Cilium?

 

Cilium provides networking, security and observability for cloud-native workloads. It replaces traditional dataplane approaches with an eBPF-based dataplane that scales with your cluster and gives you identity-aware policies that allow you to route traffic, load balance services and enforce policy with a lower overhead.

 

Why Tetragon?


Tetragon is a security observability and runtime enforcement tool. It uses eBPF to observe process execution, file access, network activity and more at the kernel level so you can detect and respond to threats without heavy agents inside every application container.

 

What about Isovalent?

Isovalent created Cilium and Tetragon and was acquired by Cisco in 2024. Cilium and Tetragon are open-source CNCF projects, and Isovalent Enterprise adds commercial features, support, and hardened builds on top of them. This lab uses the open-source versions, so everything you see here is available in Isovalent Enterprise as well.

 

What You Will Do
 

  • Start from a KinD cluster that has no CNI, then install Cilium to provide Kubernetes networking and gain visibility and control of the environment.
  • Deploy an Ollama and prompt-client AI workload so you have steady east-west traffic to inspect.
  • Install and enable Hubble UI and use it for flow visibility of the AI application.
  • Segment the inference API with a Cilium Network Policy that only allows prompt-client to reach Ollama, and observe the result in Hubble.
  • Observe and enforce system-level controls with a Tetragon TracingPolicy.

 

Lab diagram

Loading

Technologies