The traffic for uploading storage points is doubled or even tripled

I'm running Nextcloud Hub II (23.0.1 RC1) with Backup 1.0.4.
The remote server (where the storage point should be uploaded) is Nextcloud 21.0.7.

Roughly monitoring the transfer of the 100MB backup chunks reveals, that Data is transfered from NC 23 to the remote NC server that should store the backups. So far so good.
After upload is finished, traffic monitor shows the same duration of traffic in the other direction. Once that is completed, the next part goes through the same upload->download procedure.

Once all these individual "loops" are finished it seems that the server that should send the backup, receives everything again from the remote server...

I assume this behavior was implemented to be able to calculate checksums for the integrity checks. But for "private" DSL connections it is not a nice behavior since it blocks upload on the "receiving" side.

A possible solution might be to do the checksum calculations on the remote server...

Start by reproducing the upload of 100MB backup chunks between Nextcloud 23.0.1 and the remote Nextcloud 21.0.7 server while monitoring traffic in both directions. Trace the backup transfer and integrity-check flow to determine why each upload is followed by a matching download. Done means redundant reverse traffic is eliminated without losing backup integrity verification.