scaling backend services

[evgeni]

Today we deploy various services (Foreman/Puma, Pulp/Gunicorn) in a default configuration which usually spans a couple (~2) workers (threads, processes, doesn't matter) inside the container.
We do not expose any way to change those numbers today, so it might become a bottleneck in setups with higher load.

On the other hand, it feels more natural for container scaling to spawn more containers instead of changing configuration inside the container.