Mobile Edge Computing and the Data Processing Continuum

A rich opportunity for companies focussed on the “data continuum”

We recently wrote about network transformation and its role in delivering the value and promise of 5G (see here). Transformation of the RAN and core network are essential to increase the flexibility and versatility of the network and enable the leading uses for 5G technology.

One of the biggest new opportunities introduced by this transformation is mobile edge computing. It’s the result of several advances spanning networking and cloud computing, with 5G becoming a key piece of the puzzle. It will power a new phase of co-operation and competition across the technology spectrum and transform the notion of distinct domains in mobile and cloud. The lines will start to blur. Operators will begin to simultaneously compete and partner with hyperscale cloud providers, and operator networks will begin to more closely resemble those of cloud players.

So what’s mobile edge computing? A simple definition is that it’s a set of computing hardware and software resources, installed at the edge of a telecom operator network, that enable functions and applications to be run closer to the user and source of the data. This could be for multiple reasons: privacy, data residency requirements, performance, cost, agility and so on. With the introduction of 5G, the dependence on a centralized cloud for certain applications using mobile network connectivity is reduced. A high-bandwidth, low-latency connection between a device, machine or user and the edge cloud stands to enable a host of new uses and business models.

The edge is suddenly a focal point, but it’s also extremely nebulous. Customers will often ask “How do you define the network edge?”. The reality is that there’s no single answer because it’s entirely dependent on context — the edge is in the eye of the beholder. For some, the network edge stops at the base station. For others, it’s the corporate network or end point, be that a smartphone or sensor sitting in a field. The application or workflow dictates where the edge is, meaning there’s little point to philosophical debates about what the edge is.

This is an important distinction, as many cloud providers and silicon suppliers claim an “end-to-end” advantage. Capability in cloud-to-edge computing is being touted broadly, as data processing, analytics and artificial intelligence move closer and closer to the source of the data. Local workloads will proliferate, and performance, security and efficiency will become critically important ingredients. In a world where zettabytes of data will be created daily, not all that data can go back and forth to the cloud.

There’s a rich opportunity for companies that can take a role in the processing and movement of data across what CCS Insight calls the “data continuum”, from the cloud, through the network and down to the end-point device or machine. A company that can play competitively across the data continuum is in a strong position, particularly if it can deliver tailored solutions for different segments and varying requirements at the network edge. But suppliers increasingly need to be able to take a full-system view of where data flows to in a system and how it’s used.

This is why operators are scrambling to establish their own edge-cloud capabilities and relationships with Amazon Web Services (AWS), Google Cloud, Microsoft Azure and others. Hyperscale cloud players are building out their edge credentials — AWS Greengrass, Microsoft Azure IoT Stack and Sphere, and Google Cloud IoT Edge are good examples — and in tandem building partnerships with operators. We predicted in 2018 that the next few years would see a wave of collaboration and competition between cloud players and operators. This trend is already starting, with Microsoft and AT&T recently announcing a strategic alliance spanning cloud, artificial intelligence and 5G (see Microsoft and AT&T Team up for Cloud, AI and 5G).

Similarly, semiconductor players, notably Intel and Qualcomm, are increasingly claiming end-to-end capabilities. Qualcomm has the advantage of scale thanks to its smartphone chip platforms, and is trying to build that up in other embedded devices and the data centre. Intel is taking its existing strength in the data centre and addressing the data continuum from the network to the edge through its broad range of silicon platforms, ranging from Xeon Scalable processors and field-programmable gate arrays to Atom processors. Intel’s One API initiative is the “glue” that will give developers consistency in the portfolio, coupled with its Open Network Edge Services Software (OpenNESS). Its Foveros 3D stacking technology holds the promise of bringing vastly more processing power in smaller form factors at much lower power budgets.

This approach is critical to addressing the massive fragmentation that exists in Internet of things markets and arguably puts Intel at the forefront of the end-to-end opportunity, with an integrated hardware, software and services approach. Of course, Nvidia and a host of others are also playing at varying points across the continuum.

Success for all will be dictated not only by companies that have the scale and resources to address the data continuum, but also by those that take a collaborative stance, embrace open-source software and look beyond the market structures that have pigeonholed them in the past. Relationships spanning both enterprise and hyperscale sides of the opportunity in mobile edge computing are also crucial given the breadth of solutions that will be needed.

Edge computing, with mobile edge computing as a major part of it, along with 5G and artificial intelligence are poised to fuel innovation and disruption at a cadence not seen before.

A version of this article first appeared in FierceWireless on 10 October 2019.