What’s the difference between a containers and a virtual machine? How is a hypervisor/kernel actually implemented? Do these abstractions add runtime overhead? If so, how much, and where?
This presentation attempts to provide clear, evidence based, answers to the above questions. In this presentation, I:
- Discuss the implementations of virtual machines and containers.
- Trace the history behind VMs and containers. I argue that the historical prevalence of VMs was largely due to a failure of contemporary OS isolation mechanisms, and that this trend will reverse.
- Dispel common misconceptions about each technology.
- Create a Docker clone in C using Linux namespaces.
The “live” portion of this presentation consisted of a live demo of various Linux namespaces on my laptop, and was mostly lifted from “Containers from Scratch”, a presentation delivered at GOTO Amsterdam 2018 by Liz Rice. I’d highly recommend checking out the original talk, which is masterfully done (particularly the live code demo).
Slides available here.