ONF Connect 2019

Edge computing increasingly supports a wide variety of industries, including Oil & Gas, Consumer Products, and Retail. The traditional methodology for providing the infrastructure necessary for this edge computing has been based on many of the same technologies used for traditional server virtualization including containers and Kubernetes. Unfortunately maintaining a Kubernetes cluster requires providing customers with dedicated containerized processes in every edge point of presence, limiting network density and keeping edge provider costs high.

We believe that the ideal way to provide compute in a dense network is not to adapt technology developed for traditional servers. Our experience with our 180 PoP edge network has shown that lighter-weight forms of virtualization can consume an order of magnitude less server memory, while supporting scaling of capacity in single-digit milliseconds. This reduction in cost can make edge computing cost competitive with other forms of compute, dramatically widing the pyramid of use-cases it is suitable for, and consequently the industry’s total addressable market. Just as importantly, it can do it while improving margins, and while pushing compute closer and closer to devices, enabling an entire new catalog of consumer experiences.