Well, we divide file into blocks, each block is 4M, and files are divided into blocks before storing them in backend storage.

The metadata here including the filename, filesize, and sha256 hash of the blocks.

Our service need to assemble the blocks on the fly and pipelining them to the client.

Seems subrequests are needed to send multiple requests to the backend storage server and then assemble the response.

Is that correct?