Open Source Summit Europe + ELC Europe 2017

In Facebook's new container system we started to heavily utilize Btrfs, cgroups2 and systemd. The combination of these tools and some additional internal code allowed us to create a lightweight, fast and efficient container system that along with our schedulers allows us to deploy, migrate and manage the life cycle of different jobs on a large scale.

We would like to present an effective alternative to existing systems (e.g. Docker, Kubernetes, etc.) that will grant a broader choice to anyone involved in containers. Our solution is aimed at systems where full virtualization is not required, making more efficient usage of available hardware and enabling fast startup/shutdown times, easy debugability, and good fault tolerance while still providing proper isolation at large scales.