This comes from the real engineering problem: If there is a large file, what is the solution to read it and process it in parallel. How to consider the tradeoff for this problem.