What specifically separates this from Swift or Ceph, both of which have S3 compatible APIs and are considerably better documented/hardened/maintained + distributed?
At this early stage of the project it is hard to explain why another approach is necessary.