The year of the edge

This time about the "lowercase" edge. The edge of the network. Gartner is all over the edge. So are mobile carriers and smart building engineers. And they all are right. The era when we wanted to process everything in the cloud is over. That was silly. And fortunately never happened. Probably could not happen. I even heard ideas as crazy as mating a light switch with lights in the cloud. But probably the self driving cars made the final argument for the edge: when there are serious low latency requirements in situations that may be life threatening, even the hardest cloud proponents bend...

The edge is rational. Very rational. Saves bandwidth, cuts latency and increases reliability. Combined with the new concepts like the ICN (Information Centric Networking), the edge becomes the center of the new architecture of connected systems.

The edge and the ICN are ideal for building automation systems too. Local traffic stays local. Aggregated statistics are pushed to the cloud (near) real time, or even periodically, when operational costs matter or bandwidth is not always available. But at local premises everything works without delays and without silly points of failure like the internet connection. Or even a connection to a local controller, when every edge node has a controller built in. Utmost reliability possible.

The edge also radically reduces network load. Instead of having to carry each sensor event to a central box (which used to be the point of failure too), events are published locally, exactly where they matter. The information is also consumed and processed locally (or within a couple of hops at most). It is really minority of traffic that has to be pushed long distance. And the long distance traffic is very tolerant to latency.

This all makes "systems" behave in a much more relaxed, yet responsive and reliable way. Hugely increasing satisfaction of users.

Comments