Why You Should Care about Container Virtualization by James Bottomley, Parallels

Why You should Care about Containers

James Bottomley
CTO of Server Virtualization

Basically, it's all about the Cloud

Hypervisors are based on Emulating Virtual Hardware

Containers are based on Shared Operating Systems

Virtual Hardware

Operating
System

System
Libraries

Connectors

Application Support

Platform

Application

The Application represents all the utility of the virtual machine and yet it is often only a few 10s to 100s of kilobytes of code.

The Supporting machine often weighs in in the Gigabyte range.

With a container, only the actual application is encapsulated

Leaving behind all the virtual machine junk (the useless 99.9%) and leaving you with a small neat capsule.

Containers can be scaled vertically just by tuning a kernel limit. No messy balloon inflation

Containers can be scaled horizontally across nodes using tools like CRIU

Horizontal scaling (both expanding and contracting) gives you huge elasticity when you're paying by the second.

All of this makes containers much more elastic than hypervisors (i.e. much better for the cloud) simply because there's far less junk to move around

However, it is often argued that since the application is the same, anything you can do with a hypervisor, you can do with a container.

Thinking in the wrong paradigm can make some problems seem insurmountable

So how is the container paradigm different? They obviously do elasticity and density better, but what else?

Going one step further: a packaging system for containers, like docker, gives you lightweight applications with instant portability.

Once you have portable packaging, you can do application dependency and composition.

Why does this matter?

Currently there are no "Cloud Only" applications. They all could run within a desktop at a pinch.

Choose a paradigm that frees you up

Instead of one that ties you down

Presented using impress.js by Bartek Szopka

Thank You!

Questions?