How hard is it to implement a simple 2-segment strategy with the first segment from host A being downloaded normally, and if another host having the same file is found (host B) the second segment is being downloaded 'reversed', meaning it starts at the end and goes towards the start to meet up with the first segment?

Jammet