Getting DOD Linked - How to Build Netcentric Operations

摘要

Conventional theory for large networks has undergone a radical revolution, Under the old regime, networks were modeled as static, random, and passive entities. Today, we recognize large networks are actually dynamic, scale free, and competitive in nature. Why does this shift in network theory affect the DOD's goal for netcentric operations. First, if large enterprise networks such as integration of all of DOD's command and control systems are seen as dynamic, growing organisms, then we should rethink our current approach of a master, all encompassing architectural definition of the 'to be' state. Instead, we ought to capture the essential aspects of this 'to be' architecture with modest efforts and expect to evolve it frequently over time as the enterprise grows and changes. This empowers our architects to produce meaningful guides today without having to promise the omniscient vision for all time- something that is not only impossible but restraining progress in moving out today. Second, the realization that the network is not random is perhaps the biggest impact that can help us achieve netcentric operations today. Current theory states that any very large network is actually scale free or scale invariant. In a random network, we would view the number of links per node of a network to follow a Poisson or Guassian distribution. Using this old model we find resiliency to be quite robust; analysis would indicate up to 10% of the nodes could be taken out randomly and the functionality would basically remain intact. However, conventional wisdom tells us that hackers can bring down the network with a much smaller fraction of nodes being attacked. Though they probably haven't studied the new network theory, they empirically understand that not all nodes are equal- in fact some are enormously more important than others. These major hubs are the lifeline of the large network and if attacked can cripple the overall effectiveness of a network very rapidly.