Edge computing: the what, the where, and the how we get there
Once a novel concept, edge computing architecture is the essential ingredient for leveraging Internet of Things (IoT) devices and data analytics, artificial intelligence (AI) and machine learning (ML) to improve their operations.
Once a novel concept, edge computing has matured into the mainstream as a practical, potentially game-changing option for enterprise IT. Major cloud services vendors are deploying more edge servers in local markets, alongside telecommunication carrier investments in 5G deployments. For enterprises, edge computing architecture is the essential ingredient for leveraging Internet of Things (IoT) devices and data analytics, artificial intelligence (AI) and machine learning (ML) to improve their operations.
During the inaugural 2020 Future of Technology Series, KeyBanc Capital Markets (KBCM) brought together leaders in edge computing to discuss where the sector is headed. Moderated by KBCM analyst Brandon Nispel, the panel included Dane Walther, SVP, Media and Carrier Division, Akamai; Andy Stewart, President and CEO, Evoque Data Center Solutions; and Cole Crawford, Founder and CEO, Vapor IO.
The evolving definition of the edge
As Vapor IO’s Cole Crawford put it, edge computing is very much a network issue requiring an entirely new approach to internet infrastructure. As more content and services are delivered over the internet, congestion has become a problem—especially as the pandemic drives skyrocketing demand for online services. Edge infrastructure helps solve the problem of latency, the lag that occurs when data travels long distances from its source to the end-user.
Edge computing capabilities will be required to support data intensive IoT innovations such as autonomous cars and drone delivery services. “To ease the congestion of the backbone internet that we have now, edge is very much an evolving journey for everybody,” said Crawford.
The definition of the edge will continue to vary, as will the use cases, added Akamai’s Dane Walther. In the early days of media streaming, providers would deliver content from one or two central data centers. Now, the Akamai team works “nonstop” to place more servers in more data centers to keep customers’ internet backbones from become constrained.
“We want to deliver performance-sensitive or scale-sensitive products as close to the edge as possible to get the best performance,” Walther explained. “A flash card for a game download, or a new release of a movie that is suddenly online because everyone is at home—for that size of release it’s not only important to be at the edge for scale and ability to deliver the performance that people want, but you wouldn’t be able to deliver that at the speed or quality from further away.”
Of course, closing the latency gap involves physical limitations of networking infrastructure. “As smart as the modern-day developer is, you don’t defeat the speed of light,” observed Crawford. “We are bound by physics by the 75 seconds of latency required for a token to work.”
Digital transformation and the edge
Over the past five to 10 years, mid-sized and large enterprises have increasingly shifted from heavily centralized to decentralized data centers, as well as adopting hybrid computing, according to Stewart. Digital transformation initiatives are increasing the shift.
“What we’re seeing is going from a single data center in one region to now having deployments of workloads across multiple regions, multiple geographies and globally,” said Stewart. “That’s what’s happening with the enterprise, even to companies that you wouldn’t consider to be an edge play.”
Crawford considered digital transformation and edge computing from the perspective of the pandemic, such as the technical limitations for data-sharing among life sciences companies. “It’s very easy to take a megabyte of data, like a DNA sequence, and add protein folding and epigenic data processing and turn it into multiple terabytes,” he explained. “There are profound implications for the creation of networks and the demand for more small power cell deployments and a direct link to fiber.”
The real impact of the pandemic on digital transformation is not evident—but it is happening, proposed Walther. “Necessity is the mother of invention,” he said. “People are being more creative about how they’re using our services and asking for advice. What we’ve seen to date is incremental. We haven’t seen the real transformation that this will bring to businesses.”
The economics of the edge
In the traditional world of large, centralized data centers, the economies of scale, shared infrastructure and a neutral host environment, clearly lead to return on capital for investors. Where the edge is concerned, the economics are less clear as enterprises pursue their strategies in the evolving edge ecosystem.
If a video streaming company, for example, wants to deploy content delivery points on the edge to improve streaming performance for consumers, it would need to broker agreements with multiple different telecommunications companies. What enterprises need, suggested Crawford, is a means of accessing multiple points with one agreement.
For that reason, 5G networks and edge computing are inextricably linked. Both are poised to significantly improve the performance of applications and enable huge amounts of real-time data processing. However, 5G networks are being rolled out slowly, where edge computing demands and capabilities are here today.
“The fact of the matter is, 5G needs the edge, but the edge doesn’t need 5G,” said Crawford. In Crawford’s view, mobile network operators and telecommunications companies need to work together to eliminate circuitous data routing, increase route diversity and increase network interconnectivity. “Telex pioneered that business, and what’s needed is a 21st-century version of that capability,” he said.
Stewart emphasized the intricate ecosystem required for edge computing. “It is incredibly complex,” he observed. “It’s easy to talk about anecdotal applications, but to make them work you need different parties working hand-in-hand. Otherwise, it’s not going to pan out—or the economics won’t work.”
Edge computing use cases
While some cloud service providers, including infrastructure- and platform-as-a-service providers, have expanded aspects of the edge, content providers and e-commerce companies have been among the early adopters using the edge to advance their businesses.
“Forward-thinking companies have been pushing our road maps of features and services,” said Walther. “We’ve had some media customers that are really forward-thinking and made some really great decisions about how they wanted to deliver their content, and the flexibility and scale that they knew they should build in from day one.”
Crawford sees gaming, augmented or virtual reality applications and artificial intelligence (AI) applications as key use cases for the edge. The edge also is critical to support the inference engines that make AI and machine learning applications so powerful.
“Inferencing applies across multiple industry verticals and can happen in every use case—in the automotive sector, in industry 4.0, in precision agriculture, security,” cited Crawford as examples. “Inferencing is a big deal at the edge because it is really latency-centric.”
Crawford predicted rapid growth in edge computing. “This is a hockey stick,” he emphasized. “We’re at this inflection point and over the next six to 12 months, you’re going to see a lot of uptick in use cases.”
Stewart observed that the economics of the edge are changing. “A lot of the deep learning might be core today, but I do see more AI/ML decision-making be pushed closer to the edge,” he said. “You have all this data from different devices—smart homes, smart cities or things that 5G will enable—and all that data is great. But to monetize it, we’re going to have to push out more learning and intelligence into the edge to analyze it in real time and make decisions on it.”
Discussing the maturity of edge computing, Stewart argued that edge technologies are more advanced than is generally perceived. According to Walther, the development of the edge is more like a cricket match than a baseball game.
“It’s going to go a lot longer than nine innings,” he quipped. “It’s hard for me to put a number on it. The concept of edge computing is going to continue to evolve as long as people are doing business online.”
Following the baseball metaphor, Crawford views the industry as being at the top of the second inning. “Companies like ours have put multiple millions of dollars into 36 markets,” he explained. “Tons of companies are putting big bets on the infrastructure, but there are wide variations among companies that are far into the journey and those that are still thinking through the nascent opportunities at the edge.”
The future of edge computing
Asked for their bold predictions about how the edge would evolve, Crawford suggested that within five years the “edge” will simply be an extension of the network rather than a distinct component of networking.
Stewart predicted a split between companies that are active owners and operators of networking assets versus those that play other roles in edge computing and investment. “In the last few years you’ve seen—whether it’s AT&T and Verizon or some of your data center operators and technology companies—sell off assets and become pure-play companies. As you start seeing the promise of the edge, a few folks will make a run at being a one-stop shop. I’m not saying they’ll succeed, but some will convince themselves that they can do everything from core all the way to the edge and we’ll see some acquisitions as they try to put that together.”
Walther cited Microsoft as a company that potentially could be looking to create a one-stop shop and make a big move in private networks. “That could provide an angle for companies other than carriers to get in,” he said. “I’d still bet on the carriers, because they understand the complexity of building out the core and end-to-end aspect of the network."
“Metaswitch and Azure can enable any company to become a virtual mobile network operator,” said Crawford. “I do believe that will happen, and I do believe we’ll see an end-to-end walled garden [of private mobile networks].”
What is clear is that the demands for connectivity and accessibility will continue to grow. “They’re just going to continue to grow,” said Stewart. “At times they’ll grow faster and at times they’ll grow slower, but the fact we can count on is that the need to go deeper is always going to be there.”
Winners and losers at the edge
As the edge ecosystem evolves alongside 5G infrastructure, some market participants will emerge ahead of the pack and others will fall behind. Cellphone tower operators will win, predicted Crawford, along with data center operators and content delivery networks, while companies with limited network access will lose.
Walther predicted that companies that enable others to bring services to the edge will be big winners, while those not already moving their businesses online are going to be losers. “There’s a middle category of those with the potential to be winners, including carriers, but they have to figure it out,” he added. “Telecommunications carriers have a huge opportunity and it will be interesting to see how that plays out.”
Stewart declined to define winners or losers but agreed with Walther that carriers have an opportunity. “If they do it right, they can really be winners, especially in connection with 5G,” he explained. “As Cole said, 5G needs the edge—but there’s a chance that they won’t [capture the opportunity] and there’s enough opportunity for somebody else to step in.”
To learn more about edge computing’s intersection with your business or other technology trends, connect with your investment banker.
About the Future of Technology Series
This inaugural seven-week virtual series centered on timely and unique content across the Technology landscape. Leaders from today’s most relevant technology names gathered to take a thematic approach toward exploring market shifts, particularly in light of an increase in digital acceleration, remote work, deglobalization and changing consumer consumption habits. Attendees included 977 investors, 164 private and public companies, with 115 fireside chats and panels, and more than 700 one-on-one meetings. For consideration to participate in the 2021 Emerging Technology Summit, please reach out to Corporate Access to discuss.