CoreOS has emerged over the course of 2014 to become an interesting approach to building and deploying a Linux distribution, focused on container deployment.
Helping lead the development of CoreOS is CTO Brandon Philips. In a video interview with ServerWatch, Philips explains how the key components of CoreOS, including Fleet and etcd, come together and how the Linux distribution works.
Prior to helping found CoreOS, Philips had been at Rackspace, where he came to a number of realizations about building Linux server distributions. One of them is that many organizations needed to make a lot of customizations to Linux in order to get their applications to run.
"So we starting thinking about the idea of how to improve what a Linux distro looks like for people that are building Internet applications," Philips said.
Using containers as a base, the CoreOS team looked to see what else was needed for a scale-out Internet application deployment. Philips said it's important to have a system that can schedule work across machines.
That's where etcd fits in. Philips said that etcd solves the need for having a consistent global configuration for a cluster of servers.
The next piece in CoreOS is fleet. Fleet combines both the scheduling piece along with execution.
"Fleet was the natural step after etcd," Philips said. "I have a consistent data store, now I want to take my application and make sure it's running across the machines, even if one or two machines go down."
CoreOS Works Well with Google Kubernetes and Apache Mesos
While CoreOS can stand on its own, it can also work well with Google's Kubernetes and the Apache Mesos efforts. Philips explained that Kubernetes combines scheduling and execution as well as a service discovery solution. Kubernetes actually uses the CoreOS etcd peice, and CoreOS developers actively contribute to Kubernetes as well.
"Really with CoreOS all we're trying to do is provide a platform to run things on," Philips said.
Watch the full video interview with CoreOS CTO Brandon Philips below: