I was watching one upload closely (750 MB) and the amount of traffic is actually around 15x the file size.
- Every batch of chunks get uploaded x5
- at the same time same amount of data is downloaded (why?)
- Verification, every chunk gets downloaded x5
Tx/Rx are switched in the picture
Is the verification of all chunks done twice?
Is there a reason for downloading all the data to do the verification? My thought is it should be enough for the client to generate random nonce and then both client and node do hash of chunk+nonce and compare.