Presentation given at AutoMacon on September 16, 2015 in Portland, OR. This talk covered how to build CoreOS components on Debian using Ansible and a brief demonstration of running an ELKStack application on the new cluster.
Our goal is to begin a conversation across a set of Helion teams.
The topic is deeply technical involving engineering early, (and not just PM).
History:
ATG was asked to form an opinion regarding CoreOS
Conclusion: neither CoreOS, nor Project Atomic were ideal.
#1 – Create a proof-of-concept in the form of a reusable prototype + demo as a means of discovery.
#2 – Present the results to the appropriate audience(s).
The discoveries do not fit neatly into existing team(s).
It is unclear which product management team(s) could be responsible.
The issues are sufficiently technical to require both eng & pm engagement.
This presentation is the beginning of the discussion.
A note about the code name: CoreOS DNA
It is intentionally a code name.
The term is NOT intended to be a product name.
The “product” may not even be a single Helion product.
Flannel solves this by giving us networking between containers on that bridged network
This diagram depicts “as deployed” networking.
* Homogeneous – all the same
The “demo” environment depicts todays typical configuration
and is instructional.
Terminology, today, can be ambiguous.
Use the subnet prefix when reading this presentation and the demo to disambiguate.
Docker-Networking projects are rapidly evolving.
It is not clear if the proof-of-concept’s architecture will remain.
The proof-of-concept is flexible to major changes in the Docker-Networking best practices.