Technology | Edge computing

Understanding life at the edge

Words by Martin Kelman
Understanding life at the edge

To understand cloud and edge computing we need to understand what current networking capability is and what future networking capability will be.

Over the years, there’s been many names for edge computing from fog, to hybrid, even cloudlets have been used to describe it. Each of these descriptions have tried to tackle different use cases and different use cases have required different terminologies. The reality is that a lot of those use cases are based on two things; latency and bandwidth.

What is edge computing?

The guiding principle of edge computing is: the closer the computer is to the device, the lower the latency and the greater the bandwidth. Latency is the delay between the transmission of information between the source and the destination. Bandwidth is a term that describes how much data can travel along the pipeline at any one time.

In essence, information can travel at the speed of light so there shouldn’t be an issue. But today’s networks were never designed for so many connections and so much information travelling at fast speeds.

How will 5G affect edge computing?

The ever growing demand of the Internet of Things has meant more information more systems more devices and needing to connect to the cloud, this is the most common use case for edge compute today.

What if we had networks that were fast enough and could handle the data requirements of all those new devices? This is what 5G, the 5th generation of mobile networking promises. If we do have a network capable of delivering the information to the cloud within milliseconds do we need edge computing?

Where is edge computing still relevant?

One interesting example is autonomous vehicles that won’t be connected continuously to the cloud. These vehicles will need to act in an independent manner, processing information on the fly. For this use case (and many others) we need edge computing.

We will need to be able to deliver models from the cloud to the edge and update those models based on continuous improvement of monitoring of all the autonomous cars connected to the network.

Do we still need edge computing?

There is definitely the need for edge computing for today and tomorrow. The use cases may change and evolve but we’re always going to need a way of linking devices to the cloud. The next phase will be how do devices talk to each other, but that’s a topic for another day.