Keynote: A Cloud Native Swiss Knife
summary
Kubernetes log streaming by label removes node-level complexity, and ephemeral debug containers (stable in 1.25) lets operators attach a debugging image into a running pod to run tcpdump, htop or vi without bloating base images. Local development can bridge into clusters via telepresence or bridge to Kubernetes, while heavy-image pain points at CERN (examples: 19 GB images, one compressed to 76 GB / 125 GB uncompressed) get mitigated with Dragonfly peer distribution and containerd remote snapshotter. GitOps with Flux or Argo CD restores clusters in ~30 minutes after a mass deletion.
AI-written summary · about
chapters (11)
- 0:00Cloud-native Swiss knife concept
- 1:34Swiss knife analogy evolution
- 2:58Logging and metrics tools
- 5:01Debugging with Kubernetes
- 7:50Hot reloading and local dev
- 8:53Large container image challenges
- 11:41Handling massive images at scale
- 13:04Image distribution solutions
- 14:05GitOps at CERN
- 15:10Incident recovery with GitOps
- 16:34Closing remarks
related talks
17mKeynote: Kubernetes: Finally...A True Cloud Platform
07mKeynote: Container-Native Dev-and-ops Experience: It's Getting Easier, Fast.
15mKeynote: Kubernetes
20mKeynote
20mKeynote: Kubernetes - Don't Stop Believin'
08mKeynote: Cloud Native in the Enterprise: Production, Best Practice and Innovation
06mKeynote: Red Hat: Making Containers Boring (again)
17mKeynote: Nine lessons from Kubernetes in Enterprise