Regular users are supposed to obtain data from DBnomics using the user interface, the web API or a module of a supported programming language (R, Python, for the moment).
Advanced users, or application developers, may want to obtain converted data DBnomics stores in Git repositories to leverage the ability to synchronize new data regularly, downloading only what changed, as Git is designed to do like so.
Using Git protocols
GitLab allows to clone repositories via HTTPS or SSH.
Roughly, when cloning a repository, Git walks the objects graph to find the reachable objects, compresses them in a pack, and sends the pack over the wire.
These steps can be quite long with large repositories (> 5 Gb, perhaps).
So DBnomics team tried some workarounds explained below.