I know it supports s3. It also basically tries to be a an object store too with its filesystem support. Regardless of the storage backend, its object storage in some form. Seems like an extra layer to uploads that is not needed. Why not just speak s3?
The only interesting bit is that it can mirror data across different blossom servers and proxy requests... none of that works yet
The worst part of blossom is that you have to set an expiration date for files. Hopefully other implementations do it differently and thats not part of the blossom spec.