Image Source

“It’s the edges of the maps that fascinate …” — David Mitchell

Imagine you are Mark Watney stuck on Mars for five years with barely any food left to sustain your existence. You decide to start growing crops indoors without soil and sunlight (somehow) to have something to eat. Wouldn’t it be great if you had some sensors installed that could help you analyze and track the usage of water, the condition of the soil and crops? You could then determine the optimal harvest time and chemically intervene in case of deteriorating conditions. With such data in hands, you could almost guarantee you’d always have something of quality to eat.

These sensors with useful insights are an example of what edge computing is all about.

Edge Computing is computing done at or near the source of data itself, as opposed to the cloud where a central data center (or centers) does all the work. In other words, it is about bringing computing to the network’s edge.

Why bother with that, though? And why are businesses increasingly focusing on the edge computing architecture?

The thing is — there’s just too much data in today’s growing world and there are too many devices connected to the internet.

Hence, traditional computing architecture (including the cloud) built around sending all that data to a central place for storage, processing and analysis (client-server relationship) can get severely strained. Some of the issues typically encountered in traditional architectures are:

  • bandwidth limitations,
  • latency problems,
  • possible network disruptions.


If you can’t get the data closer to the data center, get the data center closer to the data. (Source)

In edge computing, storage and servers are at or near the data itself. Processing and analysis is done locally and only the results are sent to a data center for review or follow-up analytics.

The most popular example here is self-driving cars. Autonomous vehicles must “communicate” with one another to avoid crashes. Just imagine what would happen if that communication took place through the cloud or any other data center — the probability of a crash greatly increases if there are latency problems or unpredictable network disruptions. Hence, at least some (or maybe most?) of the processing and analysis has to be done on locally deployed IoT devices / sensors.

Here are further use cases of edge computing:

Computer Science & Data Analytics Master Student @ADA & @GWU