From one edge to another

Edge networking should not be confused with edge computing, according to Kris Beevers, CEO at NS1

Edge is a much used but quite ill-defined term. At its origin, edge computing described the ability to move dynamic computation farther out to the edges of the internet—closer to users and machines that relied on that compute and processing power. But as the concept of edge has evolved, it has become quite complex, now encompassing all aspects of connectivity, which creates some confusion.

While edge computing focuses on placing processing closer to users, the concept of edge networking concerns routing data and network traffic in a more optimised way across distributed footprints. The outcome is the same—lower latency and greater resiliency, but edge networking is quickly gaining steam because it gives companies more control and allows them to use existing infrastructure. Understanding the difference between the two empowers IT teams to best identify, implement, and manage the infrastructure approach best suited for their needs.

The push for edge computing was driven by the need to improve application performance and optimise server resources, particularly in the age of IoT and 5G. The first to jump into the fray were telecom providers who already had the footprint necessary to support it. Adding micro data centres and working with cloud providers enabled these companies to bring processing from centralised on-prem or cloud data centres out to the edge.

But this came with challenges. Building applications that exist across a highly distributed footprint is different from building one that operates on 1-2 data centres. Few tools existed to make this scalable and repeatable. Over the last few years, containers have made edge more accessible. Many companies now run their workloads across globally distributed Kubernetes clusters—the nodes that run containerised applications—or across serverless functions inside service provider environments.

5G will be the ultimate example of edge computing. With full roll out, edge antennas could have so much power that companies can have duplicate services running on each of them. However the use cases for this remain few. Autonomous vehicles being one example where processing and traffic to and from the node must be lightning fast to avoid accidents.

For edge networking, the possible applications appear vast. For industries such as healthcare, the finely tuned latency advantages of an edge network enable medical diagnostic tools to work quickly to produce results, speeding up treatment and diagnosis; to allow healthcare IoT devices to generate and process vast volumes of data; and to support network speeds so doctors can engage remotely with patients without fear of internet interruptions or lag time.

In many cases, the industry leaders who have built their own “edge networks'' are prime examples of edge networking. Major content providers, such as Netflix, or collaboration companies like DropBox have edge networks supported by an omnicloud infrastructure, and a great deal of investment is made into intelligently orchestrating their application traffic and automating the lifecycles of the resources that support it. They can spin up and control capacity close to audiences to ensure consistent, superior user experiences. The shaving off of milliseconds and nanoseconds is also prompting gaming start-ups to raise tens of millions so they can build highly performant edge networks that use software-defined networking that intelligently steers traffic on a packet-by-packet basis.

While edge networking may have its roots in edge computing, it is difficult to ignore the differences in implementation and delivery. A solid understanding of both will benefit IT leaders as new trends, such as the need to support remote work, push businesses to consider improvements in connectivity for both employees and customers. Fortunately, investment in this space continues to grow, and innovation is strongly focused on new platforms that provide foundational services to enable IT teams to build and manage highly distributed, edge-enabled networks and applications.