problem: find the smallest diff between a file on the server and client, so you can send only the change. rsync does this!
Decades ago, a sliding window was hashed, moving byte by byte. This was very slow. (but can use many cores!)
The naive approach of comparing 8k chunks fails if someone added a single new byte at the beginning. This is called the boundary shift problem.
Content defined chunking (CDC) attempts to describe chunks by their content,