With the original storage engine it was easy to organize incremental or differential backups, as the old files weren't tend to change. We also could choose which collection to backup, as one file meant one collection.
With the wire tiger we need much more backup space, even if one byte is save somewhere, many files are affected and so hashes are changed...

Today we have discovered, that we have 560 copies of the same heavy file, and probably having deduplication functionality on gridfs is possible, or even easier then thinking of filesystem deduplication, that could also reduce the RAM amount mongo uses to cache the data from storage, so we want to ask to add that functionality. Thanks a lot,
at TUMO Center for Creative technologies we love you very much.