Guest blog by Rajesh Ghai, Research Director, Carrier Network Infrastructure, IDC

Every day, more applications migrate to the cloud. Corporate networks increasingly stream rich data such as video, and real-time services such as 5G and IoT drive more low-latency connections and bandwidth demands. With all this activity — which seemingly grows exponentially — no wonder the economic business models of Communication Service Provider (CSP) networks are straining under the demand.

CSPs face a conundrum: Bandwidth is a finite resource and network architectures have focused primarily on maximizing the utilization of this increasingly scarce commodity. As bandwidth demand keeps growing, CSP budgets can’t keep pace accordingly. What can CSPs do to alleviate the issue?

There is a solution: CSPs can redesign network architectures.

The key to making any network redesign effective is doing it in a way that tackles the bandwith problem while changing the economics of building and operating networks at the same time. While this sounds like a daunting prospect, it’s entirely feaible. But first, CSPs have to fundamentally change the way they look at network design.

Router Capex, Network Capacity
Figure 1: A Dollar in Router Capex Supports 14 Times More Network Capacity in 2018 Than It Did in 2008

CSPs have historically managed to keep pace with the explosion of bandwidth demand mainly through advances in network silicon capacity. Thanks to Moore’s Law, such advances happened at regular intervals for decades. Indeed, $1 of routing capex today supports 14x more deployed network capacity than it did 10 years ago (Figure 1).

However, even Moore’s Law has its limits and today the networking industry’s ability to keep up is wavering. Both speed increases and lead times just can’t keep pace with demand, which is making the issue of bandwidth scarcity a bigger problem for CSPs.

Today’s network designs have sought to address this bandwidth scarcity with a variety of network design choices that often create new problems in the form of inefficiencies and operating complexity. As a result, operating expenses have taken a hit. The internet architecture based on existing design points has essentially tapped out — its shortcomings are becoming more pronounced as the internet continues to scale.

What if the fundamental design constraints of networking could be eliminated? What if bandwidth scarcity no longer dictated how networks are designed? What if CSPs could focus instead on rearchitecting networks to first and foremost simplify and reduce operations at massive scale?

Such a redesign starts by challenging set notions across all layers of the networking stack, much like hyper-scalers have done by reinventing datacenter network design. For CSPs this means alleviating constraints around network capacity by adopting a consistent and standardized design across silicon, software and systems for all network elements, and tethering these network systems to cloud-based automation engines. These network architectures will be less complex to manage and operate and more efficient to scale.

By embracing hyper-scale cloud principles to automate, operate and manage the network, CSPs can fundamentally alter internet economics.

Call to Action

  • IDC believes the CSP industry needs to move past the bandwidth scarcity constraint and elevate the design of its network architecture to focus on operational efficiency through simplicity and consistency of network element design across all layers of the networking stack.
  • For a deeper dive into how hyper-scale datacenter network design principles can be used in Internet redesign, please read the IDC white paper: Reimagining Communication Service Provider Architecture.