Images, Containers and VMs: Put Them into a Blender
summary
The talk presents a proof-of-concept system that treats Docker images and VM templates interchangeably by running one lightweight VM per Docker image and exposing Docker endpoints through a proxy VM. It maps Docker lifecycle operations (pull, run) to launching these container-VMs, demonstrating integration with orchestration (Kubernetes support planned) while highlighting tradeoffs: much slower startup (10–12 seconds vs ~20 ms) and significantly higher resource overhead.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 3:30Traditional vs cloud-native
- 5:41Cloud-native deployment models
- 7:30vSphere Integrated Containers
- 11:21Container limits on legacy ops
- 15:55Virtual Container Host solution
- 18:07ContainerVM deep dive
- 23:05Container tech adoption phases
- 27:43Container limits for legacy apps
- 31:47Docker packaging and deploy
- 34:35Focus on Docker
- 37:44Stage 3: containerize infra templates
- 40:19Self-service with ContainerVM
- 41:21Future outlook
related talks
34mThe beauty of virtual infrastructure with K8s and containers
49mContainers without Docker
31mVMs + Containers = The Perfect Wedding
45mContainers are dead. Long live virtual machines!
48mContainerize everything!
42mVirtual Machine Orchestration in Kubernetes Utilizing KubeVirt
25mIntroduction to Bootable Containers
05mBare Metal to Containers to K8s: How We Got Here