Not sure this will help in your situation but this is what we do. This applies to OSX only.
I store all documents (hundreds of thousands) in their native format outside the data file in a directory on the same drive as the 4D datafile. I store the files by client / patient in their own folder with several subfolders. I use Execute on Server to retrieve / save the documents. I store all the relevant metadata including the relative path of the document in the 4D datafile. I like to keep the 4D datafile as small as reasonable. e.g. for repairs etc.
Now for the backup, I have a separate RAID drive attached to the 4D Server. I use Carbon Copy Cloner to clone the documents folder every 15 minutes. At night, the 4D Server runs the backup to a separate directory on this second drive.
Once the 4D backup is complete, I use Retrospect to backup encrypted to a removable drive that goes offsite and to the cloud.