- In a recent survey of 14 edge thought leaders and market movers, the top edge application was video content delivery, which 92 percent of respondents identified as an area of impact and concern.
- The following three additional edge apps were each named by 83 percent of the experts: autonomous vehicles, including trucks and drones; augmented reality and virtual reality, each with its own set of applications; and industrial Internet of Things (IIoT), which includes various forms of factory automation. Rounding out the top five was gaming with an artificial reality (AR) or virtual reality (VR) user experience, which was named by 75 percent of the experts surveyed.
- Telcos enjoy an advantage in edge deployment over both internet content providers (ICPs) and over-the-top (OTT) operators. Telco edge sites are located within 5 to 20 milliseconds of the targets they serve, while the edge sites of ICPs and OTT operators are located at 20 milliseconds to 50 milliseconds.
The emerging edge is the new and powerful set of locations for handling applications. An edge can be anywhere – for example, a connected car is an edge on wheels. With so many target locations possible, some will be built to handle less data for fewer users, while others will be more massive, including data centers distributed around a metropolitan area. New low-power and specialized chips will power edge devices. While many people equate the edge with 5G, the edge and edge apps already exist today, and there have been very few 5G deployments to date.
As the IoT universe expands to include more devices – some 20 billion units, according to IHS Markit estimates – the sheer scale of data generated in the future will render the current paradigm of a centralized cloud untenable. This IoT edge processing saves upstream traffic and bandwidth, by turning huge amounts of raw data into both immediately usable decisions and directives to feedback to objects and devices downstream, and a much reduced and refined set of data to send upstream to data lakes and other decision-making systems.
Barriers to edge deployment
The main obstacle to effective edge deployment is the large bandwidth required. Of most significance is the need to reduce upstream traffic via analytics, to manipulate and refine data at the edge.
Technical considerations and high cost were also barriers to edge deployment, according to survey respondents. In fact, many edge apps require new software to be invented, yet no common application program interfaces (APIs) exist for edge app developers. For a robust edge to exist, a common developer platform is needed—so there is potential in the open-source Eclipse platform of APIs and microservices.
Also at stake is the question of how an operator can manage, monitor, operate and control thousands of edge sites, since no standard software stack exists. Another issue for automation is how to translate data from big data lakes and develop the “small data” artificial intelligence inference engines required to work at edge venues. More experimentation and experience is needed to determine suitable and appropriate business cases for the edge.
IHS Markit surveyed 14 major edge thought leaders and market movers about the most pressing issues surrounding edge applications. The survey defined the edge as 20 milliseconds from the edge compute location to the end user, device, machine or application. Note: This study was commissioned by the Linux Foundation Networking. However, the project is the sole work and responsibility of IHS Markit. The study was conducted at arm’s length, including the interviews, analysis and reporting.