Networks have become key enablers for virtually all the important business processes throughout any organization, and as businesses grow to pursue new markets, networks are a crucial part of their success. Today, the needs of service providers in particular are evolving, and a new generation of networks is arising to meet those needs — one that features unprecedented growth and change at every point in the network, from the edge to the core.
This situation presents important strategic decisions for the network operators. The decisions they make now will impact their networks for the next five years or more, including their ability to maximize revenue streams. Service providers planning a multi-service edge strategy must account for exponentially growing connections plus a diversity of service use cases, each with its own list of specialized requirements.
The multi-service edge: Diverse demands
The rise of edge computing has given rise to plenty of jokes about the risks of “living on the edge,” but in fact, the challenges created by edge computing are no joke. Futuristic services involving AR and VR are no longer in the future. For example, one in five adults in the U.S. have tried VR, and 70% of VR users plan to increase their usage in the coming year. The global IoT market to $158 billion. And work from home is here to stay — one projection has 25% of all professional jobs in America remote by the end of 2022, with a further increase in 2023.
To maximize revenue, service providers must have the flexibility to support all of the above use cases and more. From a technical perspective, dynamic, multi-service edge nodes that can serve both consumer broadband and business functions require high logical scale. They must be optimized to support complex features for each service, meaning very large route forwarding tables, flexible tunnel encapsulation, rich QoS, security filters and the ability to attach traffic management counters to each.
The core: Higher throughput
The growth in business demand will not only increase the diversity of services to be provided. It will also increase traffic — dramatically. For example, the International Telecommunication Union (ITU) estimates that overall mobile data traffic will grow at an annual rate of roughly 55% in 2020-2030, reaching 5,016 exabytes in 2030. This level of growth will put tremendous pressure on other parts of the network, like forwarding nodes for transport and core. Core nodes don’t have to host the same diversity of network functions, but they will need to process huge amounts of traffic to keep up with growing throughput demands. As a result, they need to be optimized for bandwidth throughput and forwarding performance.
5G: Technology-driven requirements
The arrival of 5G further complicates the picture, adding technology-driven requirements to those driven by business needs. Its combination of speed, bandwidth and ultra-low latency will not only enable consumers to have a better video experience on their mobile phones (which large numbers of consumers will demand), 5G will also be an important B2B play. Factory floor automation, IoT and eventually autonomous vehicles will be major markets, and the most important ones may not yet even be known. In other words, 5G will add to the challenges throughout the network.
The trade-off dilemma
The need for high logical scale at the edge and high throughput at the core creates an engineering dilemma all the way down to the silicon level — and it needs to be resolved if service providers are to meet their business goals. The specialized application-specific integrated circuit (ASIC) chips that ultimately power network platform can be optimized for flexibility and logical scale, or they can be optimized for bandwidth and throughput — but not for both. Unfortunately, logical scale and throughput are diametrically opposed variables. Within a given die size and power budget, an ASIC can optimize for superlative bandwidth throughput or superlative logical scale. But one can’t increase without decreasing the other. Attempts to reach a compromise between these two conflicting needs can result in less-than-optimal performance everywhere.
Not the time to compromise
End users couldn’t care less about the ASICs that power the switches that control the networks they depend on. What they do care about is their experience, and that is directly related to the architectural choices networking executives are now facing. To meet short- and medium-term growth objectives, service providers need a network that gives them flexibility at the multi-service edge and speed at the core, and this requires different platforms that are purpose-built for these differing tasks. Compromises may be technically possible, but service providers must have the right platform for the right task to be confident that their network can meet growing demands — including ones that can’t be predicted in advance.