Gartner defines edge computing as “a part of a distributed computing topology in which information processing is located close to the edge—where things and people produce or consume that information.”

The 2021 State of the Edge report from Linux Foundationfindings describe the edge to be a natural extension of cloud computing and a key enabler for the "Fourth Industrial Revolution" – largely equating to closer to IoT devices. Albeit a natural extension, is there a new breed of applications which need the edge?

The genesis of the edge is alongside "latency critical" applications and data which have local importance. With a growing number of IoT devices, it does make sense to deploy edge on manufacturing floors, retails stores, agricultural IoT frameworks and other control-system defined use cases. Do we refer to this as a Deep Edge – where there is a defined footprint from functionality, manageability, and price points?

Perhaps there is an aggregated edge – referring to the edge which serves “latency sensitive applications” and is also specialized to handle certain workloads which have GPU requirement. We call this where AI meets the edge. As hyperscale cloud providers embrace edge and telecom service providers (5G) distributing edge via MEC, workload behavior extends beyond the local manufacturing floor construct.

Edge taxonomy should include regional edge, beyond current footprint brought to us by hyperscalers. The characteristics of this edge should purely be defined by application requirement with a few axes:

1. Latency
2. Quality of Experience (QoE)
3. Data residency / Sovereignty
4. Location based on security posture

While edge connotation is changing, SaaS solutions and enterprise applications with centralized deployments in cloud or in data centers would need such a strategic edge in order to support the customer base – keeping in mind the axes described above.

There are many use cases which can utilize data processing capability, video processing capability, and AI at the edge. Inferencing at the edge has a greater impact than traditional execution in the cloud. Beyond inferencing, information security and privacy have several use cases which can utilize edge for data processing. Simply put, there lies an edge where actionable data must be processed where it is generated. As more and more products are offered as SaaS solutions, dis-aggregating workloads closer to the customer will lead to enhanced QoE and help shepherd implementation of data residency. This is done while simultaneously saving processing time at a much smaller set than processing zeta bytes at the central lakes.