Crises in network evolution: three case studies and one proposed solution

Three case studies are used to illustrate the network evolution process. Similarities are extracted showing that network evolution is a series of crises. A crisis is characterized as the inability of a network or its supporting organization to handle its workload because of lack of resources. The case studies also show that the crises follow a natural ordering which allows for their prediction. A new approach called executable models is proposed to help smooth the evolution process. The first case study describes an experimental installation of an integrated voice-data office switching and communication system. The second case study emphasizes the dynamic management problems facing a campuswide cluster of networks installed at Boston University. The third case study deals with the ARPANET. The main conclusion is that one has to emphasize the design for testability through the entire system life cycle