If the Jeopardy answer is “Cloud, Fog, or Edge,” the question is “Where all can you do analytics?”
We rejoice in the PROFINET of Things with its Data Access, Uptime, and Openness, but the Industrial Internet of Things is really about analytics. PROFINET is an enabler. A very important enabler, but still an enabler. The central theme of the IIoT is to save money by improving the process and preventing downtime by analyzing data from the production system.
But where should the data go to be analyzed? The early IIoT answer was always the cloud. Because: Big Data. The cloud certainly has some advantages. The database and software are maintained by others – it’s outsourcing this function. My analogy is hobby websites. If you were early to this field, you might have hosted your website on a computer in your house. (Like you did the BBS you started with… for those of you acquainted with ancient history.) Now you will most likely have your website hosted professionally. There are hundreds of companies offering this service. They handle software updates, connectivity, and uptime. Likewise, the Cloud for Big Data.
The cloud is not always the answer today. Here are some of the disadvantages as reported in the Computer magazine article, Fog Computing for Smart Living.
Cloud-only models face serious challenges in latency, network bandwidth, geographic focus, reliability, and security. Fog computing reduces these challenges by providing a system-level horizontal architecture to distribute computing, storage, control, and networking resources and services from the cloud to connected devices (“things”). Think of fog computing as the cloud on the ground: it enables latency-sensitive computing to be performed in close proximity to the things it controls.
You may have guessed that this was penned by a proponent of fog computing, but the points are still valid.
Do the engineering to determine which of these approaches is best for your project. As I said in a previous post, Low-lying Clouds, there’s never a substitute for doing the engineering.