And ANSSI Says: Here’s How to Secure a Container
summary
Applies ANSSI Docker guidance to a VM-per-VM Docker Compose setup: enforces memory and CPU limits, assigns distinct cgroups per container, centralizes logs to Logstash via the GELF driver, isolates networks per service, and drops most kernel capabilities using cap_drop all (effective for ~90% of services). Replaces docker-in-docker with kaniko for image builds, notes read-only root and per-container user namespaces with bind mounts are impractical, and concludes the changes greatly reduce attack surface though builds and application-level security still need further work.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 3:10Docker context and use
- 7:23Container resource limits
- 10:54Filesystem security
- 16:51Docker log management
- 18:37Network security
- 22:27Namespaces isolation
- 27:46Kernel capability reduction
- 30:10Rules in practice
- 37:54Build security and conclusion
- 40:41Q&A
related talks
33mDocker Security
14mYour Containers Aren’t Alone: Demystifying Container Isolation
47mContainer Security
35mDocker Security
31mYour Containers Aren’t Alone: Demystifying Container Isolation
33mContainers Security Layers and How Not to Break them
46mSecuring Applications and Pipelines on a Container Platform
16mSecurity and Kubernetes: Should We Go Back to VMs?
