It works exacly as it was designed to work.. GIT as VCS.. Version Control System.. for text code sniplets. It can handle small binary blobs just fine.

If you need (D)VFS aka Distributed Versioned Filesystem, grab right tool. Or write one.

This is exacly way I wrote DOT (Distributed Object Tracker). Its pure DVFS repo manager, to handle binary blobs and that it.. Nothing more.

People complaining about GIT not working well w/ big data just handling GIT wrong. Linus said it from the begining, its NOT tool for such datasets. Just move along.