5G: New edge builders needed
November 26 2019
by Brian Partridge, Raymond Huo
5G will alter the role of wireless connectivity in society (in many places eliminating the need for fixed broadband connections) and provide a central platform enabling the hyperconnected world. While the transitions from 2G to 3G and 3G to 4G were largely driven by consumer demand for more capable mobile data service, 5G addresses an unabating consumer demand for faster broadband and delivers enterprise functionality beyond broadband access to the internet for IoT and low-latency/mission-critical communications. The fully realized 5G vision will see multiple virtual networks running over a single physical deployment of RAN, edge cloud, middle mile and core technologies (see Figure 1).
5G is unlike any previous generational shift – never have the cloud and network been so necessarily converged. The 5G era we have now entered introduces enticing new B2X revenue opportunities, as well as a new set of risks that span technology, business processes and culture. Important questions loom about the viability and business case for 5G services, especially for services or use cases where 5G is billed as disrupting the status quo, such as for industrial enterprises or where operators enter new value chains, such as working directly with application developers. In the long run, 5G will require multi-access edge computing to deliver on the latency performance of sub-10, -five or even -one millisecond, dictating that a new buildout of compute capacity is required. Who will build and operate such infrastructure?
Figure 1: 5G is three virtual networks in one deployment
The 451 Take
Vertically integrated 5G systems unique to each discrete operator would appear both uneconomical and inferior to shared infrastructure models often seen in the cloud and datacenter sectors. Enterprise leaders will demand edge cloud services that bring multi-cloud services, connectivity hubs and capacity for their own IT assets. It would seem unlikely that any single operator could cost-effectively monetize such architecture on their own, save for a select few. If the existing 5G ecosystem doesn't stand up this new shared infrastructure, then MNOs could, once again, find themselves disintermediated by the combination of private network builds and private edge computing. We think that constructive industry dialog on this topic is required now, ahead of the key functionalities that will come with 3GPP Releases 16 and 17 over the next few years.
5G: Promising and different
Today, many of the world's mobile devices are running on some form of 4G (LTE) network, which was an evolution from 3G, and 3G was an evolution from 2G, and so on. The previous generational shifts brought significant new capabilities (such as mobile data services, short message service and mobile broadband) and underlying architecture changes (from analog to IP), but 5G brings change to a new level of complexity and capability, with software-defined networking dynamically routing data based on quality-of-service (QoS) needs and available communication channels.
Multi-access edge computing (MEC) will bring the distributed computing power to deliver ultra-low-latency performance for B2X applications. This infrastructure, once deployed, opens new opportunities for operators to engage directly with application and managed service developers by offering wholesale or retail access to this computing platform far closer to users and devices than the public clouds operating around the world today.
When will 5G be ready to fulfill its promises? This year and next can be viewed as the initial phase of 5G rollouts focused on enhanced mobile broadband for smartphones and fixed-wireless access services. Beginning in 2021/22, the full feature set of 5G should be standardized and ready for primetime. Important questions must still be answered on how 5G networks are ultimately built, by whom and for whom.
Are telcos heading down wrong path?
Tier one/top 20 operators, having the current course and speed for building and monetizing edge computing, look to leverage a model that is vertically integrated. That is, they will build their own edge footprint for optimal control and differentiation. For instance, tier one operators like AT&T and Verizon have signaled plans to deploy hundreds of edge nodes in various locations across the United States, leveraging their existing real estate holdings to transform points of presence such as central offices and existing datacenters into edge capacity (see Figure 2). We are presently concerned that this track will serve them well for only a subset of edge opportunities – namely the ability to serve their own vertically integrated network applications over a geographically distributed footprint. This approach will largely miss the mark for enterprises that might have considered using these edge clouds as part of their IT stacks or looked to retain options to leverage multiple commercial or private 5G networks. What enterprises will require are the services we most often equate with datacenter companies and REITs, which offer local breakout offramps to popular public cloud services, connectivity hubs, and compute and storage capacity for rent in any number of commercial offerings.
Figure 2: A fully realized 5G network's distributed topology
Network operators, save for a small handful, have not achieved much success in attracting third parties to work deep within their networks and IT infrastructure. Instead, they typically aggregate access to such entities at relatively few interconnection points, to minimize cost and protect their service and governance models. We expect that industrial enterprises will eschew public network options that fail to meet their requirements for multi-cloud edge connectivity, advanced edge services and ecosystem-friendly connectivity. Shared asset models have worked in the past in the telecom ecosystem in areas like passive RAN and backhaul sharing, neutral hosts, and radio towers. We think that the industry at large should consider these proven, economical models when it comes to edge computing. The counterargument we see brewing is: If we share, how will we differentiate? The answer, as it always must be, will come down to holistic customer experience, service experiences and deft technical skills that transform the 5G promise into business outcome reality, especially for industrial B2B segments.
5G: A new edge infrastructure investment opportunity?
5G has the power to revolutionize many parts of today's society. It's disruptive technology that will only fulfill its potential if disruptive approaches to building it are taken. Highly reliable low-latency communication services can bring fresh capability unlike anything possible in previous wireless generations, which rightly has the attention of leaders in the manufacturing, transportation, warehousing/logistics and healthcare industries. 5G should not be viewed as a 'last mile' solution only, but also as a 'middle mile' and 'core' infrastructure that can be harnessed in support of industrial enterprise digital transformation.
This means the intermediate connection networks between 5G RAN and 5G CORE can be 'sliced' and consumed on a discrete service basis. In theory, this could enable a smoother transition to multi-cloud setups by tapping off-premises IT via consumable edge cloud infrastructure, enabling new disaster recovery, data sovereignty and resiliency options. For greatest economic efficiency, this edge cloud could host 5G-specific virtual network functions alongside virtualized, enterprise-owned IT functions, and then be orchestrated together to create a multiplicity of unique services tailored to an individual customer's IT needs for the moment. These types of services could then be dynamically modified over time, either by human intervention or automation, to accommodate changing business needs and enable business transformations. All this context leads to our punchline: We think fresh investment and new actors are needed. Let's consider the United States as an example.
If a datacenter REIT player like Equinix or Digital Realty were to build out a 5G-optimized edge footprint (say 250-300 geographically relevant locations), it could actively serve the three or four major MNOs, MSOs, rural and minor MNOs, MVNOs, private/semi-private enterprise network builders, and smart cities and municipalities, in addition to supporting hyperscale cloud edge services. This new multi-tenant, hyperconnected edge cloud footprint could inject extreme efficiency into the entire ecosystem, mimicking the operating models and efficiency of well-trodden models seen all over the world (RAN sharing, wholesale IP/backhaul, tower sharing, etc.). This new buildout would also accelerate 5G ecosystem development in countries where it is adopted, helping achieve geopolitical objectives in areas like Industry 4.0, autonomous vehicles and remote healthcare, among others.
We believe the current course and trajectory for 5G as it relates to edge computing is destined to underwhelm B2B stakeholders. New actors should be brought to the table from the datacenter/ICT investors, government/regulatory bodies, REITs, MTDCs, tower companies and the MNOs themselves, to immediately start meaningful dialog on shared infrastructure models that can serve the greater good most efficiency and economically. We suspect the outcome after careful study of the issue might ultimately lead to opportunities for new or existing ICT actors to enter the 5G service delivery space, or more aptly, serve the 5G ecosystem and
the edge cloud ecosystem.