Clearing the Confusion: Defining the Layers of Edge Computing

Dean Bubley, a futurist at Disruptive Analysis, looked at latency in edge computing at the recent Datacloud Europe in Monaco. (Photo: Rich Miller)

This is the first of a two-part series on the State of Edge Computing. Part One looks at the future scale of edge infrastructure, and the challenges in defining and funding its different layers. Part Two will explore the curation and analysis of edge data, and how it will influence industry business models and the design and placement of infrastructure.

Edge computing will have a transformational impact on American society and business, enabling new technologies and services across instant wireless connectivity. On that point, there is widespread agreement.

But first, edge computing must navigate a maze of confusion and dueling definitions, fueled in part by the diverse nature of edge infrastructure and business models.

The good news: major stakeholders in edge computing are taking steps to clarify the various infrastructure layers and use cases, which is a key step in laying the groundwork for success. Edge computing is not a single “right” solution, but rather a number of technologies that will work together. There will be a range of niches where businesses can specialize and succeed.

But challenges remain. The timing remains unclear on when the business demand for the different flavors of edge infrastructure will materialize, complicating decisions on how and when to fund these projects.

It’s been six months since our last deep dive into the State of Edge Computing. During that time, Data Center Frontier has held a series of discussions with leading players in the edge computing sector, seeking to provide readers with clarity on a trend that is both buzzy and fuzzy. Our take: There has been significant progress in clarifying the layers of edge computing and how they can enable new businesses.

Driven By New Technology

Edge computing is not a single technology, but a phrase that describes several layers of infrastructure, some of which are refinements of existing models. Edge infrastructure is a response to new technologies – such as autonomous vehicles and distributed AI applications – that require low latency and close proximity to users. These technologies are shaping the future of Internet infrastructure,

The transition to edge computing will be an enormous undertaking. It will take years, cost billions and billions of dollars, and ultimately become ubiquitous in scale.

“What we’re trying to do is remake the world, said Josh Snowhorn, Chief Strategy Officer at EdgeMicro, a provider of edge colocation services.

It’s a task that includes the telecom infrastructure for the rollout of the new low-latency 5G wireless network, as well as data center infrastructure to house servers and storage hardware for applications – as well as the analytics to enable a new layer of intelligent, real-time services.

“This is the first time a network has been rebuilt since the Internet,” said Cliff Kane, founder and Co-CEO of Cleareon Fiber Networks. “You’re going to need a lot of (edge sites), all over the place, deployed quickly. There will be millions and millions of small cell sites. All that fiber has to go back to the edge data center. You’ll need to reroute the traffic back to the constituency it serves.

“We believe this is the future, and want to be the plumbers and landlords for this infrastructure,” said Kane.

“This is the first time a network has been rebuilt since the Internet. You’re going to need a lot of sites, all over the place, deployed quickly.”
Cliff Kane, Cleareon

“All of this requires a connected infrastructure,” said Phill Lawson-Shanks, Chief Innovation Officer at EdgeConneX. “The edge is going to be of every size and shape. There’s an infrastructure deficit that we need to address. It really does need to be everywhere.”

Neat Definitions Prove Elusive

For edge computing to be deployed everywhere at scale, it will need to be compact, cheaper and take the network to new places. The key components of edge computing include mobile devices, wireless networks, telecom towers, small cells and distributed antenna systems (DAS) as well as data centers and cloud platforms. It’s a universe that transcends traditional business models built on specialization.

That’s why there’s so much confusion about edge computing, according to Dean Bubley, a futurist at Disruptive Analysis, who chaired the edge computing track at Datacloud Europe in Monaco.

“The data center industry thinks edge is a module,” said Bubley. “The small cell industry thinks edge is a small cell. Device makers think analytics will run on sensors.

“Edge has become this umbrella term,” Bubley added. “The data center industry speaks some of the language of the 5G community, but you’ll see different terms for edge.”

Mark Thiele of Ericsson moderates a discussion at the DCD-Edge conference in New York in May. (Photo: Rich Miller)

“Edge to one person may be core to someone else,” said Hugh Carspecken, the CEO of DartPoints, which provides modules for edge computing. “There is no standard deployment.”

As with cloud computing, it will take time for the market to grasp a holistic understanding of edge and its many layers. Meanwhile, some vendors and analysts offer visions that focus on a partial picture or specific solution. Industry veterans believe there is progress being made, due in part to the scope of the business opportunity.

“There is no ‘edge war’ to be had,” said Rob Nash-Boulden, Director of Data Centers at the engineering firm Black & Veatch. “No one needs to arm wrestle over this. This is going to help solve our problems. This isn’t an either-or proposition. It’s all happening simultaneously. It will be an evolution, not a revolution.”

“There is no ‘edge war’ to be had. No one needs to arm wrestle over this. It will be an evolution, not a revolution.”
Rob Nash-Boulden, Black & Veatch.

“No one company offers a one-stop solution you can go buy,” said Mark Thiele, who leads the edge computing team at Ericsson. “There’s no such thing. Finding partners, maybe co-opetition, is the key to smoothing the path.”

Glossary, Reports Seek to Clear Confusion

In an early example of this, a group of stakeholders has teamed with The Linux Foundation to create the Open Glossary of Edge Computing, along with a companion report that outlines broad principles of edge computing. Participants include VaporIO, Packet, ARM Holdings, Ericsson and Rafay Systems.

In a parallel effort, Vertiv has spent more than a year developing an overview of edge use cases, summarized in a report (Edge Archetypes) that outlines its definition the layers of edge architecture.

“As edge computing continues to evolve and expand, our goal is to bring clarity and simplicity to the critical infrastructure required to support the future of edge computing,” said Martin Olsen, vice president, global edge and integrated solutions at Vertiv.

An overview from Vertiv of the layers of edge computing infrastructure. Click for a larger version. (Source: Vertiv)

Vertiv sees five layers of infrastructure:

Data Sources: Sensors, things, VR headsets
Processing Endpoints: Wearables, cars, robots or PCs
Local Data Hubs: Data storage or processing at a small IT room or building
Urban Data Hubs: Data storage or processing at a cell tower, metro data center or event street furniture.
Regional Data Hubs: Cloud availability zones (AZs) and colocation facilities

“It’s very much a core-to-edge story,” said Olsen. “As devices are added on the Internet of Things, more data will be generated at the edge. Not all of that data will go to the cloud or core. At the edge, you make decisions. At the fog (local and urban data hubs), you do deduction. At the cloud or the core, you do your machine learning.”

The report from the Linux Foundation consortium describes the emerging edge as “an exciting time for boring infrastructure.”

“Edge computing now has its Rosetta Stone,” said Matt Trifiro, Chief Marketing Officer of VaporIO, an edge infrastructure specialist and a lead contributor to the the glossary. “We originally created the glossary for our internal use, but as we shared it for review, we were asked to make it a community resource.”

“We are witnessing the convergence of many forces driving edge computing, from the advent of super-powerful GPUs to the business imperatives driving network upgrades,” the report says. “Bringing these forces together will create the opportunity for new, complex, and unprecedented edge applications. … Like most new ecosystems at early stages of development, the excitement and potential of edge computing is a complex set of definitions by a wide range of participants, adding to the confusion around this topic.”

The Linux Foundation group offers a chart of edge layers that tracks fairly closely to the Vertiv report instructure, albeit using different terminology. Here’s a look at a diagram outlining their vision:

The report also outlines four principles that frame the participants’ lens on edge:

The edge is a location, not a thing;
There are lots of edges, but the “edge we care about today” is the edge of the last mile network;
This edge has two sides: an infrastructure edge and a device edge;
Compute will exist on both sides, working in coordination with the centralized cloud.

Both the Vertiv and Linux Foundation reports outline the importance of 5G wireless as a foundation of edge computing. Wireless carriers will require edge infrastructure to deploy 5G technology, making them a key customer for edge infrastructure providers. Once it is available, ultra-fast 5G connectivity will enable new services that will create demand for edge capacity.

An overview of the infrastructure transition from 4G to 5G wireless. (Source: Vertiv)

“Edge computing will unleash a cornucopia of opportunities for every size and shape of institution, from governments to grocery stores,” said the report from the Linux Foundation group. “Applications will leap from the pages of science fiction and become practical services. We will witness a major transformation of today’s internet.”

How to Pay for Edge Infrastructure?

This revolution won’t be cheap. Some executives predict the buildout of edge computing will pressure the economics of data infrastructure.

“We fundamentally have to come up with new ways to run things, or there’s no way we can afford this,” said Jim Fletcher, a strategy partner at Momenta Partners, on a panel at the DCD-Edge event in New York.

“Applications will leap from the pages of science fiction and become practical services. We will witness a major transformation of today’s internet.”
Linux Foundation report

The 5G infrastructure will be the “tip of the spear” for edge infrastructure to follow. Thus, the geography of the 5G networks and associated business revenue will guide much of the early investment.

“The big telcos are being very pragmatic about how they launch 5G,” said Michael Ortiz, Chief Development Officer at DartPoints. “They’re going to Tier 2 markets, and you have to increase the number of towers. It’s capital-intensive and it’s time intensive.”

“All the edges won’t be equal,” said Nash-Boulden. “5G is not going to be a constitutional right. If you want to find out where the fast networks will be, follow the money. If there isn’t a business case or use case, will consumers pay more for faster Internet?”

“This isn’t cheap and it takes a long time,” said Mike Hagan, CEO of EdgeMicro. “We want to help the (mobile network operators) fix their business model problem. They’re becoming entertainment companies. How are they going to monetize that investment? I’m not sure users will pay a premium to go faster.”

Disruption Creates Demand for Solutions

Demand for services has a way of making the economics work, even if there is disruption along the way

“Five years ago, Netflix was pretty much destroying the Internet,” said Lawson-Shanks. “Most of Netflix lived in Amazon, in three or four places around the world. By moving that content closer” a better solution was created.

A major part of that solution was the EdgeConneX network of 24 regional data centers, which provided local caching of all that Netflix content for network and cable operators, who are among the company’s anchor tenants and investors. As new technologies emerge, driving demand for local storage, the economics of edge will gradually become more compelling, says Lawson-Shanks.

“All of this requires a connected infrastructure, he said. “We will look back on this time and see it as the fourth industrial revolution.”

“We will look back on this time and see it as the fourth industrial revolution.”
Phill Lawson-Shanks, EdgeConneX

And that will mean data centers in new places, which has been a core thesis of Data Center Frontier. “There will be data centers in these new places, but they may be built and powered very differently than the server farms of Silicon Valley and Northern Virginia,” we wrote upon our launch in June 2015.

As edge computing takes shape, we are approaching the brink of a new phase for America’s Internet infrastructure.

“We have learned that the core compute doesn’t exist in one place anymore,” said Martin Levy, head of Network Strategy at content delivery specialist Cloudflare, in a panel at Datacloud Europe. “Everything is about to change.”

COMING TUESDAY: A deep dive into the curation and analysis of edge data, and how it will influence industry business models and the design and placement of infrastructure.

Explore the evolving world of edge computing further through Data Center Frontier’s special report series and ongoing coverage.