This book has described the details of an emerging new architecture for the Internet of Things. But new architectures rarely displace legacy systems unless there is an overarching benefit that drives their adoption. For the IoT, the major benefit can be expressed in the unique new relationships possible between the myriad end devices and the big data servers that analyze and control the data flowing to and from those end devices. Data Drives a Change Fundamentally, the coming billions of

more »

... t of Things devices will simply generate too much data to be analyzed in traditional ways. Instead of the usual one-to-one predefined IP legacy topology, only a publish/subscribe model allows the big data servers to be selective and adaptive in the choice of data to operate upon, and is thus smarter over time. Even more importantly, the big data analyzers will not even know what data streams would be useful until they discover the data. Information neighborhoods created through data stream affinities will present opportunities for selecting and combining small data flows from many different kinds of end devices, not all of which are even part of a specific application. This allows IoT applications to become smarter and smarter over time, as ever more end devices are installed (see Figure 7 -4). Whatever initial purpose these end devices serve, they may also unexpectedly and unpredictably benefit other applications that discover their data outputs and find them useful (if the chirp streams are made public). When initially installed, specific appliances, sensors, and actuators may serve a particular application. But over time, new end devices may be deployed by the same or other organizations. Data streams from these new devices may also be recognized by "affinities" of place, time, or correlation to be incorporated into the original application's information "neighborhood."