There is no container
summary
Claim: “container” is just a userland illusion over Linux kernel features and implementations differ. Details: uses namespaces and cgroups, clone and setns syscalls, overlayFS and user namespaces (landed ~three months ago), networking via iptables, sometimes replaced by KVM. Built outcome: a PaaS that eschews opaque base images in favor of immutable builds and a replicated persistent filesystem, enabling per-branch staging and full auditability. Result: works — clusters cloneable, HA and reproducible deploys.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 2:10Why containers
- 5:25Simple abstraction
- 9:40SetHostname
- 11:10Mount
- 13:10Isolate process IDs
- 17:15SystemD
- 18:10Container
- 19:20Container abstraction
- 24:15Console
- 24:45Abstraction
- 26:00Persistence
related talks
40mUnpacking the Container: A Deep Dive into Virtualized Container Technology
36mMitigating Noisy Neighbours: Advanced Container Resource Management
37mCRI: The Second Boom of Container Runtimes - Harry Zhang & Xu Wang, HyperHQ
08mContainerization – It’s Not Just for CPUs Anymore: Introducin...
37mLevel Up Your Java Container Images
43mLinux Container Performance Analysis
58mDay 2 Session 1
18mImpossible is Nothing! The 10x hpothesis
