As 5G rolls out, edge computing is the hope of service providers to finally provide differentiated services. Service provides have two paths to monetize the edge: a consumer services path and an enterprise services path. The former plays to the strengths of the mobile service providers. The latter, enterprise path, is as an incremental opportunity that 5G could provide.
There are two deployment models to provide edge computing services to enterprises. The first is on-premise services where edge computing is located at the enterprise. The second is off-premise services where edge computing is located at an edge data center. How will service providers go about monetizing the edge and which of these options they will pursue remains to be seen. Here, I will focus on the service provider opportunity for the on-premise enterprise segment leaving the off-premise and consumer segments to a future post.
Benefits of the Edge
The benefits of the edge are well understood, but it’s important to note the approach of different ecosystem players:
- Reduction of latency and its variability;
- Reduction of backhaul costs;
- Localization of data to meet security, privacy and regulatory requirements
In the context of wireless telcos, the focus is on latency [see here]. The cost of backhaul is not a top consideration because the location of edge computing in mobile networks does not help save access bandwidth which is the most expensive link. This contrasts from the approach of the Cloud Players discussed below.
The 5G Factor
The deployment of on-premise edge computing is part of a private wireless network. 5G aims to provide a latency of 1-2 msec in comparison to 20-30 msec that is possible with LTE.
There are a few architectural features that allow 5G to achieve cost and performance advantage over LTE for private wireless networks. 5G allows the enterprise private network to connect to a service provider core network ‘cleanly’. This was not possible with LTE where the standard architecture would have been ‘broken’ had edge computing been located on premise. A private LTE network would typically have its own on-premise core which adds to the cost of the network.
In short, the market uptake of private wireless networks and on-premise edge computing are intertwined. This leads us to consider spectrum and business model factors both of which will play a critical role in determining the uptake of on-premise edge computing solutions.
The Spectrum Factor
Spectrum is a primary obstacle to cellular private network deployments. This is slowly changing in some countries. In the US, the CBRS band will provide between 80-150 MHz of unlicensed spectrum. In Germany, the regulator plans to issue location-based licenses in 100 MHz of spectrum to enterprises. The UK is considering a process to allow enterprises sublease spectrum from the MNOs, which traditionally is very rare.
Spectrum conditions favor large enterprises. Small and medium size enterprises face high obstacles in adopting on-premise edge computing solutions from telcos unless the spectrum situation changes.
The Business Models
Private networks in the enterprise raise an important question: What is the role of the service provider? This role is yet to be defined across the entire chain of sales, planning and design, deployment, and maintenance and support. Equipment vendors, service providers, enterprises and system integrators will have to collaborate. The business models for the edge will be customized depending on the context. Setting up these business models is a challenge to overcome as it is not ‘native’ to many service providers.
The Approach of Cloud Players
For the Cloud players, edge services are in part an extension of cloud services that fit the hybrid-cloud model favored by many large enterprises. Here, edge computing serves three major purposes:
- Reduce the cost of data transport to the cloud;
- Allow local data processing without connectivity to the cloud;
- Meet requirements for data security, privacy and localization regulations.
The edge for Cloud players includes the end device or a local network of devices. Their solutions include access-agnostic computing and storage hardware coupled with edge software compatible with the cloud stack.
In contrast with telcos, Cloud players don’t sell edge services based on latency. Surely, local data processing reduces latency; however, the main driver is minimizing the cost of connectivity. This subtle point has commercial and market consequences. Telcos justifies edge computing by anticipating commercial viability of low-latency applications. However, many of these applications are immature with many challenges to overcome that take precedence over connectivity (e.g. content and cost of production for AR/VR). On the other hand, Cloud players are enabling IoT (local aggregation of devices, offline operation) and AI/machine learning (reducing cost of data transport, improving response time). Both IoT and AI are established cloud-based services that are being made more widely available by extending towards the edge.
Market Trials
Over the past few days, I had the opportunity to hear Huawei’s view on edge computing while attending the 10th Mobile Broadband Forum. The event provides a platform to hear from service provider CTOs about their approach to industry developments. Automation of ports and machine vision for manufacturing were featured as two examples of on-premise edge computing deployments.
In the manufacturing use case, Huawei demonstrated its edge computing solution in a private 5G network in collaboration with Haier and China Mobile. The edge computing solution included a relatively small server (4U). The server supports different hardware resources depending on need: for example, FPGAs for fast near-real time processing, GPUs for numerically intensive applications, and GPPs. Through software, operators and factories could define and provision slices of service KPIs according to different use case requirements.
Critical Success Factors
The above example highlights two critical aspects that will determine whether service providers can monetize the edge:
- Business Models defining the role service providers will play to stimulate the entire value chain including aspects such as go-to-market strategy, system integration and support.
- Providing real benefits to the enterprise based on solid business case. Enterprise are concerned with improvement in efficiency: lower production costs/COGS and higher profit margins. Edge computing has to offer a clear value proposition.
It could be a mistake for the telecom sector to focus on latency alone as it has been with 5G. The Cloud players are already aggressively rolling out products and services that help the enterprise today in improving productivity.