How can we effectively design these systems to meet our performance needs? Can widely used operating systems be used such as Windows? What software is available for users on this type of machine and how do we run this software on other machines built using the same technology? Can we use existing applications or do we need to develop new ones, if so how? How can we ensure that each PC is doing its fair share of work, or is not overloaded?
This document explores these issues from theory to practice, details a design methodology and shows by experimental investigation that from a structured design the speedup obtained with many PC’s can be known within bounds prior to implementation.