Assume of a fog computing node as a physical server that resides amongst the edge gadgets (thermostats, robots, in-auto desktops) and the back-conclude devices, normally hosted on community clouds. The fog nodes answer to an architectural problem: too significantly latency to go requests all the way back to the community cloud-based mostly products and services and not sufficient horsepower to course of action the knowledge in the edge product by itself.
This 3-tier technique adds one more compute platform that is ready to do some—if not most—of the back-conclude processing. This addresses the issue that less expensive edge gadgets with reduce energy really don’t have the processing and storage energy to offer with the incoming knowledge natively. Now knowledge can be despatched to the fog node for processing, with out the latency effect of going all the way back to the distant cloud products and services.
Although fog nodes are a very simple alternative to a very simple difficulty, you must have an understanding of when and when not to use them:
You must use fog nodes when knowledge that is intricate or in big quantities want to be processed regionally and would overwhelm the edge-based mostly product that is consuming the knowledge as well. In other phrases, you want something that responds in practically serious time, this sort of as a factory robot shutting down when the servers overheat. You want that to be instantaneous.
You must use fog nodes when a human is waiting around for the reaction to return back to the edge product. Like the want to answer to occasions that can’t wait, human beings want to be regarded as near-serious-time architectural parts. Folks are too high-priced to be waiting around about for responses from the distant cloud devices.
You must use fog nodes when you are multiplexing knowledge from a number of unique kinds of edge gadgets or sending a number of unique kinds of knowledge. The fog node manages the processing of knowledge from a number of edge gadgets at as soon as, working with issues this sort of as semantic transformation of the knowledge prior to sending it to the back-conclude cloud servers. Or the fog node can course of action and answer immediately to the edge-based mostly product.
Essentially, you must not use fog nodes except if the architecture and needs match the previously mentioned criteria. In my e-book, they must be used sparingly, in particular thinking of that they add cost and operational complexity—and one more layer that can fail.
Your very best bet for now is to exclude them, and then determine out if you have cases exactly where they can be of use.
Copyright © 2020 IDG Communications, Inc.