One of the challenges with the Edge is that its definition varies based on individual use cases and perspectives. During a panel at DCW, each panelist offered a distinct viewpoint on what qualifies as the Edge:
Mark Gusakov, advisory board member for the Nomad Futurist Foundation, defined the Edge as any point where the network connects to improve latency.
Jake Rasweiler, Senior Vice President of Product Innovation and New Platforms at American Tower, expanded on this by describing the Edge as small-footprint facilities designed to bring certain capabilities closer to users, delivering high performance and low latency.
Jeffery Martin, Vice President of Business Development at Ascent, viewed the Edge as any facility that doesn’t require a full-time staff presence but still needs to deliver the same Service Level Agreement (SLA) as larger data centers.
Hugh Carspecken, Managing Director of Ubiquity Edge, observed that the concept of the Edge has existed for nearly 50 years but is now being redefined.
“The use of the Edge is what defines it,” Carspecken explained. “Edge computing handles local requests and actions, distinguishing it from traditional hub-and-spoke models where servers might be located hundreds of miles away.”
The Edge is Evolving
Edge computing is evolving in multiple ways. Vladimir Galabov, an analyst at Omdia, pointed out that the pace of data center construction in mature markets is slowing.
“Data center buildout will continue, but hotspots will shift to emerging locations,” he said.
In addition to shifting to tier-2 municipalities, edge solutions are set to absorb some of the demand. Mark Gusakov highlighted examples from Toronto and other Canadian cities where space constraints and green belt restrictions limit large-scale data center developments. Consequently, numerous small data centers are being integrated into existing buildings to navigate challenges like sustainability laws and space limitations.
Automotive edge computing module, close-up, digital photography, vehicle data processing for enhanced navigation
Data centers
Edge Data Centers
Modular Data Centers
Prefabricated Data Centers
Mark Gusakov noted that “Organizations are placing edge data centers wherever feasible and where they are most needed. Edge can be more efficient as there is no staff on-site, reducing downtime by about 60% by eliminating human error.”
Jake Rasweiler emphasized that the Edge consists of multiple tiers. He described several categories:
Cloud Edge: Reduces latency for cloud service users.
Aggregation-Based Edge: Provides networking and critical computing resources.
User Edge: Supports use cases like single-tenant spokes connected to major hubs, on-premises point-of-sale sites, factory automation, and more.
Cloud-Heavy Edge: Addresses cloud bottlenecks for users with complex cloud workloads. For instance, one user required edge resources to manage performance across five different cloud platforms.
Rasweiler highlighted that edge computing enables standardized designs and economies of scale with small modular units. Out of 23,000 American Tower sites, only 1,000 are large and complex.
He advised the audience to understand local power constraints and consult utilities about where they can provide additional megawatts, identifying potential edge locations.
“You need a consistent design with uniform cable trays and features across each site,” Rasweiler explained. “Customers expect reliable service from edge facilities despite their smaller size, and AI helps us achieve this.”
Jake Rasweiler
Edge Data Centers Are Like McDonalds
Hugh Carspecken likened the expansion of edge data centers to McDonald’s growth alongside the American highway system.
“The Edge is like McDonald’s units situated near the highway network,” Carspecken said. As highways expanded, McDonald’s steadily added franchises near exits, paralleling how edge data centers proliferate with network expansion.
Unlike American Tower, Ubiquity Edge customizes each of its numerous edge data centers to fit the customer, site, and workloads. “It’s vital to understand the data footprint and location specifics,” Carspecken added. For example, one airline required 2,400 edge units, each with two servers, but every unit had different specifications.
The automotive industry also drives edge innovation, needing processing, storage, and relay units to support autonomous driving. Companies like Mercedes are developing business models where they earn more from selling data derived from cars and drivers than from selling the vehicles themselves.
Security is a major concern for unmanned edge units. Panelists at DCW 2024 warned IT professionals about risks such as vendors leaving new equipment deliveries unattended near remote edge boxes. Incidents of theft, including stealing copper from these sites, were also highlighted as significant security issues.
Edge Data Centers Are Like McDonalds
Maintenance of dozens, hundreds, or even thousands of edge units poses significant challenges. For example, if 2,000 modular units in the field share a relatively uniform design, a fault in one component might necessitate replacing that component across all units. Consequently, a well-structured maintenance plan is crucial. Typically, organizations engage local partners like TekClean to expedite this process.
In Ascent’s case, the company tackled edge implementation and maintenance costs by appointing a vendor manager. A key strategy in this initiative was vendor consolidation.
The Edge (Mostly) Remains Strong
While edge deployments have revealed challenges related to maintenance, customization versus standardization, security, and design, the consensus is that the edge remains a dynamic area of data center growth and IT innovation.
“Vendors continue to be bullish about the data center industry’s continued growth, which they expect will be further boosted by demand for edge capacity,” said Daniel Bizo, Uptime Institute’s Research Director.
Drew Robb Author Drew Robb has been a full-time professional writer and editor for more than twenty years. He currently works freelance for a number of IT publications.
