Editor’s note: The teams from RCR Wireless News and Enterprise IoT Insights recently convened a wide range of telecoms and enterprise stakeholders to examine all things mobile edge computing at the Mobile Edge Forum virtual event, available on demand here. This article is taken from a session at the Mobile Edge Forum.
Edge business models in practice—operator-owned and delivered as a service vs. enterprise-owned and operator managed
As 5G networks, both public and private, are scaled out by operators and integrated into enterprise technology workflows, the ambiguously titled “edge” has become a focal point for debate. The premise is clear: to fully leverage the throughput and latency of the 5G air interface to move data, apply some sort of intelligence and turn that into a decision, compute power has to be distributed out of centralized data centers to overcome the limitations of the speed of light. What’s less clear is who should own and operate what. And with many things 5G-related, the answer is, it depends.
Daniel Beazer, a senior analyst with Analysys Mason framed the issue: “The problem can actually be quite simply stated, which is that actually delivering services, be it compute, storage, streaming, CDN, what have you, from a…massive centralized cloud node or massive data center campus from a interconnect point in Northern Virginia or London or Frankfurt…is something that makes very good sense economically, but that gets a bit shakier as we go out to something that’s really distributed and a long way away from the center. So I guess the question is, who’s going to pay for it and what’s that business model?”
‘It’s a smorgasbord right now in the early days of edge computing’
Equinix specializes in digital infrastructure, particularly data centers, colocation points and other internet connection services. Service providers use Equinix’s platform and infrastructure for a range of services like VPNs, SD-WAN, security applications and 5G network deployment. Orange for example, has used Equinix’s data centers for almost 20 years to host its backbone network, and also works with the firm on expanding its telco cloud environment.
As it relates to the metro edge, Zachary Smith, Equinix’s global head of edge infrastructure services, explained the focus is “to provide an ecosystem of partners and capabilities at those locations, and make the effort and the friction less for getting the right thing in those places connected to the right network.”
He continued: “Where we think that we can get enough aggregation to make economics for what we’re going to consider to be more general purpose distributed use cases, really more economical.” With partners like Orange, Smith said, the goal is “bringing their networks and allowing local breakout and interchange of traffic to occur in those metro points instead of backhauling all to a few centralized locations.” Looking at highly latency-sensitive services in the sub-15 millisecond range, “I still think we’re a little ways off from…being able to help customers reach all of those.”
Looking at the business models currently in play in the market, Intel Senior Director of the Smart Edge Product Milan Djukic said, “There’s some situations where the enterprises are either, because of a strategic decision that they’re taking or their internal competency and skillset, where they’re more leaning towards working with a partner to purchase some business outcome as a service.” In other scenarios where the end customer “is a little bit more vertically integrated…they’re more open or willing to purchase the infrastructure themselves and deploy and manage their own various use cases.”
Calling out variances related to particular vertical segments and geographies, “It’s a smorgasbord right now in the early days of edge computing,” Djukic said. “To be successful in this type of environment, we really need to be flexible in how we package and how we deliver our solutions with our go-to-market partners.”
Back to Orange, the operator’s International Networks Transformation Director Idir Fodil said customers require both centralized and decentralized computing resources, and that having access to both architectures of telco cloud, “We can offer our customers with two…capabilities.” Edge capabilities allows Orange to adhere to business-critical SLAs regarding latency. “Being with latency less than 10 milliseconds, you cannot achieve this with centralized cloud.” This dual telco cloud operating model lets orange provide “customers with new services, onboarding them very quickly.”
Fodil said that, in his opinion, having a large footprint of smaller, distributed data centers gives “quick return on investment” compared with large, centralized cloud resources. He also referenced Orange’s collaboration with Equinix: “We are relying on the expansion of this footprint with Equinix on this end-demand PoP in order to be in capacity to quickly roll out and deploy new services” including, 5G roaming, SASE, 5G cores, voice, and other services.
Developing an as-a-service mobile edge computing offering
In terms of how end users are communicating how they want to consume edge computing and related services, Smith said a consumption-based business model tied to specific business outcomes is pushing companies who build and operate the platforms to “figure out how we can bring that value equation to customers quicker.” He also called out the importance of extending equipment lifecycle, understanding carbon footprint and recycling components where possible in business model development.
“How can we bring more as-a-service consumption models to our ecosystem…which does move the risk around,” is the first big question, he said. “And then number two, also making sure that we’re looking at all aspects of that infrastructure reuse, extend the life, find different use cases…[to] maximize on the ROI.
Djukic, discussing the challenge particularly around on-premises edge, is that “one size does not fit all…What works in industrial versus retail versus healthcare, these are extremely unique requirements. And integrating into these brownfield environments can be extremely complex and costly.” But, he said, Intel is racking up edge use cases with a demonstrable positive ROI, which allows them to easily quantify the upside on investment.
“I think there’s enough openness in the industry, whether it’s with the partners here on the panel or with Intel to come at our packaging differently to make the numbers work so that value can be extracted by the end customer.”
Following Djukic’s commentary around on-prem deployments, Beazer turned the discussion to multi-tenancy, shared edge platforms. When operator’s take this approach, he reckoned “use case stacking” is part of unlocking an attractive business case.
Continuing to use its partnership with Equinix, Fodil of Orange cataloged edge services that can be offered to multiple customers. “We have the SASE, which is very important for our B2B customers. We have the 5G—it may be private networks. We are seeing demand for the 5G and 4G local breakout…We have deployment of voice service” and the ability to legacy voice networks into an as-a-service layer that can be tailored to specific customers.
Back to the core question of who should own edge computing node, Fodil said partnering with Equinix gives Orange the ability to rapidly and flexibly deliver services to their customers. “We kind of focus our business on providing our customers with services…We partnered with Equinix to rollout all, let’s say, the down layer, which is the hardware, compute, storage, etc…We have business model and TCO that…at the same time benefit for both Equinix and Orange.”