The core components of a Kubernetes cluster are explained, including the master node, worker nodes, etcd, API server, and kubelet.
Key Kubernetes objects like pods, deployments, services, and volumes are introduced as the building blocks for configuring applications.
Namespaces are used to group resources, while authentication, authorization, and admission controllers provide granular access control through the API server.
Containers achieve isolation using Linux kernel features like namespaces and cgroups, but share the host kernel, creating a different security model than VMs.
An attacker can chain exploits, starting from an application vulnerability and escalating to a full container escape and cluster compromise.
Misconfigurations like privileged containers, disabled namespaces, and unpatched software in runtimes like runc create significant security risks.
A live demo shows how a kernel vulnerability like Dirty COW can be exploited to escape container isolation and gain root access on the host node.
The `create pod` privilege is dangerously powerful, and security can be improved by enabling hardening measures like seccomp profiles and Pod Security Admission.
Kubernetes network policies are essential for segmenting traffic, while true multi-tenancy is extremely risky and requires advanced solutions like hardened runtimes.
Despite the risks, containerization offers security advantages through small, understandable workloads that allow for tight security profiles and automated scanning.
The shared responsibility model in cloud Kubernetes services is discussed, highlighting that users must still explicitly enable many hardening features.
46:36Andrew Martin
44:00Dimitrij Klesev & Andreas Zeissner
23:08Rico Komenda
40:00Rob Richardson
28:47Mathias Tausig
28:36Noaa Barki
57:24Hannes Norbert Göring
27:52Maish Saidel-Keesing
Jobs that call for the skills explored in this talk.
smartclip Europe GmbH
Hamburg, Germany
Mittwald CM Service GmbH & Co. KG
Espelkamp, Germany
Brink Group