Someone should spend time researching how we should model the nodes. What does the average computer in a server farm look like