Computing at the edge: Let’s get on board

Edge computing is one of those terms that means different things to different people; its definition differs based on the context. The term is used widely in internet of things (IoT) environments where sensors are amassing huge amounts of data. Rather than transmit this data to a back-end server for processing, an edge computer, close to the sensors, will process the data and only transmit aggregated data or alarms to the back-end server.

Another use case is a factory. A multitude of computers control the production environment that is manufacturing product. Often, the factory will have a local network connecting systems to allow staff to monitor the production process, and for machine control systems to exchange information. This could be an Ethernet network if the factory is cabled, a simple Wi-Fi environment if latency is not an issue, or an SD-WAN for sophisticated networking with heightened security, latency and redundancy requirements.

It makes no sense to communicate all this data back to an administrative system that does not need to know how many times a drill press has operated; it just needs to know how many finished products have hit the conveyor belt.

Vehicles are an interesting use case. Anyone who has driven a late-model car gets an inkling of how much computing is occurring in the vehicle the minute an alarm goes off and the brakes slam on, just because the design engineer thinks you’re too close to the car in front. Edge computing is vital in a vehicle in order to manage communication to and from it, be it the policeman investigating an accident, a technician providing maintenance, or the traffic light you are approaching with your left turn indicator activated.

Homes are going to need their own edge computer, too. With systems controlling the lights, air-conditioning and entertainment devices, and controllers monitoring the solar panels, deciding whether to feed the batteries or the grid based on the spot price for electricity, an immense amount of processing is occurring that does not need to leave the house. Given the cost of internet traffic, whether it be fibre or 5G, an intelligent router, minimising communication on the public internet, is a good idea.  

There are many reasons why edge computing makes sense, but the two biggest are security and cost.

Edge computing can be a physical device such as a Wi-Fi router that connects a network to the internet, or cloud-based services that provide supervisory control to operational systems and aggregate data from devices on the network. This enables security controls in operational technology environments – restricting access to production computers ensures they remain dedicated to their mission-critical workloads.

Cost is also an issue. If data is being communicated to head office from a factory environment, you don’t want to send more than you have to. Most telecommunications carriers have subscriptions that charge for the data transmitted – properly managed, this can significantly reduce the communications expense for a company.

5G only amplifies this saving because it is optimised for data, it provides more bandwidth on existing infrastructure than 4G, and offers lower latency. The network slicing capability of 5G means that an edge computer is pretty much mandatory. It is now possible to run separate data streams on the same communications infrastructure, segmenting the network traffic to meet operational (speed and safety) requirements and providing services to meet administrative (security) requirements. 

An edge device can also be a viable solution for the integration conundrum: to what degree should operational technology (OT) environments be connected to information technology (IT) services? While a simplistic solution to strengthening cyber security over operational technology is to isolate the OT network from the IT network, this seriously restricts the OT environment from benefiting from corporate administrative services and it can perpetuate bad practices that often prevail on the OT side to make up for lack of integration.

For instance, in the absence of access to the corporate identity data, many OT environments must manually provision and deprovision user access entitlements. Edge computing can assist by providing access to just the pertinent data from the IT infrastructure to enable OT operations.

So, the future is bright for edge computing. It gives us better control over our increasingly complex computing environments, and it can reduce our costs. The telcos have figured this out – with declining revenue from communications, on a per-Gb basis, moving into the services space makes sense. A good telecoms engineer, who understands 5G core network functions, is best placed to optimally leverage 5G for edge computing functions.

So, edge computing is no longer just a term bandied about by smug CIOs – it’s something we all want.

Graham Williamson is a fellow analyst at KuppingerCole