By: Marc LeClerc

Multiple forces are pushing network edge architecture to become more agile, scalable elastically, and enable service providers to more efficiently and economically deliver multiple services and applications. This article describes key characteristics of this evolved network edge, and one economically advantageous approach to building such an architecture by fully integrating computing, networking and storage into an all-in-one programmable multi-access and pre-integrated platform.

First, let’s address a couple of things driving the evolution of network infrastructure and edge computing. An increasingly important proportion of today’s business activities are transacted online, often in real time. These activities often integrate elements from multiple services that are chained together to synthesize information from different sources, automate processes, or in some way provide added value for customers.

In addition, volumes of transactions can fluctuate by orders of magnitude in mere hours. In many cases, these activities require constant evolution of existing network capabilities as well as capacity, giving rise to the need for both enhanced scalability and functional agility. The complexity inherent in delivering these needs, as well as keeping service latency within acceptable ranges, incentivize the movement of service processing ever closer to the edge.

The rising economic importance of online business has also brought with it a significant increase in attempts by malevolent actors and cybercriminals to prey on it, raising the stakes for network and information security. The network edge is the first line of defense. How cybersecurity is implemented here has a major impact on the entire network capacity for risk prevention and mitigation, while significantly impacting the velocity and costs of introducing new services and business models.

And speaking of cost, it has become a major factor, as normally the deployment of edge infrastructure equipment is more likely to be distributed to branch offices, enterprise campuses and multi-tenant premises constrained by limitations in site space, power availability, and cooling capacity.

The Service Provider Case

Another factor pushing for network evolution is the arrival of 5G networking. 5G places stringent demands on the forwarding plane and user plane, impacting both network infrastructure and the application space, and forcing a much tighter integration of networking, compute and storage resources. The massively multi-subscriber nature of carrier services further highlights the need for reducing latency, network flexibility and the ability to dynamically chain services. In carrier networks, edge characteristics are often seen as key limitations to the effective deployment of VMs, and the ability to scale services up and down in order to maximize price and performance. Additional requirements include the abilities to:

1. Cope with the communications needs of many different types of devices,
2. Ensure the network’s ability to carry multiple different types of traffic with each having widely divergent characteristics, and
3. Efficiently deliver many different types of services with each imposing their own requirements on compute capacity, response time, volume of storage, and networking.

Due to the public nature of their offerings, service providers are even more concerned with the response to the ever-increasing threat of cybersecurity breaches and denial of service attacks. Traditional architectures were not designed to cope with such volumes and so are hard to scale. Rising complexity at the edge is thus becoming a limiting factor in protecting the network core, as well as increasing constraints on deployment and onboarding of new services. It also significantly augments traffic latencies and has become an impediment to service scalability.

When you look at the overall evolution of fixed and mobile networks, their user planes are essentially coming together. We have also seen a trend toward the disaggregation of network functions, where services are delivered on hardware that provides a software-agnostic platform instead of the traditional purpose-built appliances or edge devices. 

On the left-hand side of Figure 1 (next page), you typically see radio-access devices, customer premise equipment or multiple CPEs, as well as other IoT devices, which are connected via the aggregation side typically using lower capacity links between 1Gbps