What Docker Users Can Learn From Erlang

Dmitry Demeshchuk
Senior Erlang Engineer at Machine Zone

Current ways of utilizing Docker or other Linux containers suffer from reliability problems: containers die, fail to communicate between each other, can't easily save or recover internal state, etc.

The goal of this talk is to apply Erlang's best practices to running Linux containers, making it possible to assemble reliable systems from unreliable components. Turns out, learning from Erlang not only helps to improve the systems' overall stability but also suggests solutions to some other common problems: in-container configuration, inter-container communication and so on.

Talk objectives:

* Projecting Erlang VM and its internals onto Linux host system running containers, figuring out their common things and differences.

* Borrowing Erlang's methods of improving systems' reliability: messaging, supervisors, monitors, process naming, etc.

* Suggesting potential directions where container-based clouds can go.

Target audience:

Engineers interested in using Linux containers, cloud users and developers.


Erlang happened to be Dmitry’s first production-level language, about 7 years ago, and it's still his favorite.

Prior to joining Machine Zone, Dmitry used to program Erlang at Mochi Media and Echo (formerly known as JS-Kit).

Twitter: @demeshchuk

Github: doubleyou

Back to conference page