The End of the Cloud ... prologue

Exploring my new DeepLens device this week I was reminded of Peter Levine’s thesis in 2016 that a reversion from centralized Cloud Computing back to the Edge was about to commence. Two points of interest for me are how this will disrupt the current mass migration to a Cloud-centric enterprise and how to model such an environment, especially in the realm of Security.

On the first point, Clayton Christensen posits in the Innovator’s Dilemma that the only way for market leaders to survive disruptive technology is to disrupt themselves before others do. This is clearly the approach Amazon is following by distributing a Machine Learning video camera which processes models in the device. Though I manage the device mostly through the AWS dashboard, ML processing occurs in the camera once a template has been deployed.

Thinking about how to model such an environment reminds me of the early days of my career working on the Strategic Defense Initiative (SDI), where a constellation of satellites communicated with each other in real time, inverting the Cloud metaphor with periodic downlinks to the Earth. Each node was connected to the others in a fabric that passed along local events and allowed extrapolation of larger system wide phenomena.

With cars as computing platforms we see this same approach, as the current Vehicle to Cloud model evolves to include communication between cars and their surroundings including other cars, traffic signs, government services etc., culminating in a V2X model.

Security in such an environment then, is encapsulated both within each node and the connectors between them in the network, with certain types of information rippling through the fabric of connected devices.  The participants in the conversation would change continuously as the vehicle moves through space, with differing levels of Security required based on the attributes of the services provided.

Coming back to my DeepLink camera, once I understand how to program something useful in it, I’ll look to connect it to my long ignored AWS IoT Button, forming a small static network, with a modicum of independent thinking.