Did you visit any of this year’s major communication conferences? If so, I’m sure you noted service providers and vendors talking a lot about the increasing relevance of the network edge. There’s a common understanding that powerful edge servers will soon be hosting virtual network functions and control processes in proximity to the customer or an enterprise appliance. But what might look like a logical complement to central clouds needs to be seen as a disruptive change to long-standing network design best practices, with complexity being centralized to create economies of scale and operational savings. So what created this shift in network architectural preferences and how will this unfold next year?
What services will move to edge?
There are two main reasons for putting more functionality at the edge of the network. Firstly, moving data processing from the core to the edge improves latency, scalability and availability, as well as minimizing backhaul bandwidth. And secondly, uCPE presents a very attractive business case for the outsourcing of enterprise IT services that are already provided by hardware appliances on the customer premises. That’s why 2019 will see the moderate deployment of uCPE use cases snowball into mass rollout. In addition, there will be a major push for edge-hosted software appliances driven by open communities such as TIP with vRAN and distributed cell site gateway (DCSG) projects in mobile networks.
Who will win the race to the edge, communication or cloud service providers?
While network operators already have geographically dispersed networks supported by skilled services teams, most cloud service providers build their business on hosting and storage in central data centers. That’s why communication service provider might have an advantage in rapidly adding functionality to the edge of their network, especially if they also have in-house system integration competence. Cloud service providers are investing in open and easy to deploy extensions of central could services. Initial offerings such as Azure IoT Edge or AWS IoT are continuously being improved for ease of operation and extended with AI-components for creating additional value. This is an interesting opportunity for enterprises with strong IT teams capable of integrating and operating their own IT infrastructure.
Is complexity at the network edge compromising security?
Complexity is the enemy of information security. As simple network devices at the edge of the network are replaced by multi-technology edge devices with powerful servers, the attack surface significantly increases. On the other hand, the ability to host security applications at the edge adds additional protective controls for higher security of both the communication network and user data. Besides the hosting of additional security applications, edge compute should securely authenticate with the network, flexibly encrypt user and control traffic at any network layer, and be designed in a tamper-resistant way. Building on those capabilities, a communication service provider can improve network security and develop new secure services for trustworthy networking – a service quality with increasingly significant value.
What other markets can be addressed with powerful edge compute solutions?
In 2019, innovative service providers will exploit the potential of edge computing with two key use cases: uCPE und virtualized cell site gateways. While uCPE is addressing business customers, it’s likely that the mass deployment will soon extend into residential markets with smart home gateways, as well as into enterprises with IoT gateway applications. Over time, these gateway applications will overtake uCPE and mobile edge solutions in number. Those emerging high-volume edge gateway markets will become the battleground for established service providers and cloud service providers, and both might join forces with specialized IT system integrators.
How will the design of edge nodes evolve?
With the introduction of uCPE use cases, a simple demarcation device for bandwidth services was replaced by a server for virtual connectivity demarcation in combination with hosting of virtual network functions. The next step will involve cost optimization. Depending on performance requirements at the edge, the functions provided by the edge device will be implemented in software on standard servers, assisted by physical network functions that can offload processor-hungry functions from software into hardware. High-bandwidth forwarding will be implemented with standard switches in combination with network operating systems. We’ll see a significant variety of edge nodes in 2019 with different hardware capability, server capacity and forwarding performance. However, this huge diversity won’t result in increased operational complexity. SDN abstraction and open control interfaces will ensure that economics are optimized at the cost-sensitive, high-volume network edge.