Wednesday, March 4, 2015

The Cloudcast #180 - Understanding CoreOS Distributed Architecture



Topic 1 - We’ve had you and Brandon on the show before, and I’ll be the first to admit that I didn’t quite understand what you were doing. Then the concept of an auto-updating (remotely managed), security-centric, container-centric LinuxOS started to make sense. And now it’s much broader. Can we dig into your architectural philosophy on all these projects?

Topic 2 - You mentioned early on that security was very important to you. Walk us through how you think about security differently, especially since you and Brandon both used to be Ops people when you sold your company to Rackspace.

Topic 3 - Help us understand etcd, since it’s used by CoreOS and also many other popular services. How is a scalable K/V store managing database inputs as well as things like service discovery, replication and networking?

Topic 4 - Today we have things like Mesos and Kubernetes, which can manage/schedule large numbers of resources. And they can work with various container-level units. Are we going to look at things like Docker or Rocket the same way in a year? Will they still be DataCenter-level managers, or do you expect other services/tools to expand?

Topic 5 - What’s a typical operational model with your customers? How do they manage the separation between them deploying and managing applications and CoreOS managing the OS updates?

Topic 6 - How quickly are you seeing application models changing to all the new things CoreOS (and others) and bringing to the underlying resources? Do the developers understand the “new possible” or are they pushing you guys?