What is edge computing?

Edge computing, as its name suggests, refers to data processing that's done at the edge of a network, near the source of the data. 

In a network, data is typically sent from devices be that computers, smartphones or assembly-line robots, back to a central data centre for processing and analysis. For example, a robotic arm will send back reports on how many gaskets it had picked up and moved onto a conveyor belt, or to give a more mundane example, apps on your phone may relay performance data back to the developer's data centre (or cloud service) for mass analysis.

The endpoints and data centre in a traditional network may be located quite far apart, potentially even on different continents. Historically, this hasn't posed a problem; the "velocity" aspect of big data's "three vs" refers to the torrents of data pouring into the data centre, rather than the speed with which it needs to be analysed and acted upon. Increasingly, however, connected devices - such as Internet of Things (IoT) gadgets - require instantaneous, or near-instantaneous feedback.

In these situations, the latency involved in sending data to and from the data centre is too great, even if it's analysed and turned around immediately. This is where edge computing comes in.

Data processing and analysis in edge computing can be done in a small, on-site data centre (such as a micro data centre) or, increasingly, in the device itself. 

Real life examples of edge computing

Oil rigs are a good example of how edge computing is used in the real world. Because they are typically based on remote offshore locations, they rely on the technology to mitigate lengthy distances to the data centre and poor network connections, and it's also costly, inefficient and time-consuming for rigs to send real-time data to a centralised cloud. Having a localised data processing facility helps a rig to run without delay or interruption

Similarly, autonomous vehicles, which operate with low connectivity, need real-time data analysis to navigate roads. Gateways hosted within the vehicle can aggregate data from other vehicles, traffic signals, GPS devices, proximity sensors, onboard control units and cloud applications, and can process and analyse this information locally.

Security risks at the edge

As an extension of a data centre, edge computing infrastructure naturally increases the surface area exposed to threats. Many edge computing devices aren't built with traditional IT security protocols, meaning that unsecured endpoints could be roped into distributed denial of service (DDoS) attacks, and could even offer hackers access to the wider network they connect to.

Physical security also needs to be a consideration, if devices are accessible to bad actors or people who could tamper with them.

However, if appropriate precautions are taken, edge computing can actually reduce IoT security and privacy risks by limiting the data flow between the collection point and the core storage centre, according to research from IDC.

What next for edge computing?

According to Gartner's "Digital Business Will Push Infrastructures to the Edge" report, data generated and processed by enterprises outside of a traditional data centre will increase from less than 10% in 2018 to 75% by 2022.

This is hardly surprising given the increasing popularity of IoT devices among both businesses and consumers, and while it's unlikely our roads will be full of autonomous vehicles any time soon, we still need this type of technology to operate properly.

The analyst house has also predicted in its "Hype Cycle for Emerging Technologies 2019" report that additional edge technologies, notably AI and analytics, will come to play a key role in this technology in the next five to ten years.