Originally posted in Converge! Network Digest on December 9, 2015

Spurred by increasing competition from over-the-top (OTT) service providers like Apple, Facebook, Google, and Microsoft, many network operators have an urgent need to transition and adapt their networks in order to be able to quickly develop, deploy and monetize competing services as well as to open their networks to new services that may originate from outside of the operator. This need for ‘service agility’ for network operators is being enabled by the increasing maturity of Network Function Virtualization (NFV) technology whereby many network nodes will increasingly be virtualized. In other words, over time mobile network cores will look more like the datacenters you would find at the OTT service providers, than more traditional central offices with many racks of purpose-built equipment.

While practically every network operator has virtualization as a key criteria in evaluating core network element, most current deployments are of legacy physical platforms consisting of both hardware and software. AT&T’s Domain 2.0, Deutsche Telecom’s Terrestream, and Telefónica’s UNICA efforts are prominent examples of network operators aggressively driving this change through their supply chain. While we are at the very early stages of IT and datacenter technology spreading to operator core networks, we believe that this architectural transition is generational and will gradually increase over the next decade.

Those network elements that are low-bandwidth and compute-intensive can be moved into a virtualized datacenter, either in an operator’s network or a public cloud network, more easily than data intensive network functions that benefit from specific packet processing or transcoding. This change offers operators several advantages like cost savings, scale and efficiency. Deploying common, off-the-shelf (COTS) server hardware and virtualization software is potentially less expensive than purchasing dedicated hardware for each network node or function. System capacity can also be scaled up or down almost instantly depending on demand. Redundancy and high-availability features are often inherent in datacenter infrastructure and would not have to be duplicated, as is the case in many network functions today. As service providers begin to trial and evaluate NFV architectures and individual virtual network functions (VNFs), this process will likely trigger another round of vendor selection for network elements centered on the NFV and management and network orchestration (MANO) architectures.

Given this change, the value of the funds spent on networking infrastructure is slowly being reallocated. Network specific hardware spending is trending downward—toward zero perhaps—while these funds will ostensibly be reallocated to software elements that perform the same function on top of COTS hardware. These software elements will come at a lower price than the hardware-based networking solutions. Moreover, the software elements, or VNFs, will be sold in the traditional manner of perpetual licenses with an annual software maintenance fee, but also as an annual or multi year subscription aka networking-as-a-service (NaaS) or infrastructure-as-a-service (IaaS). Service providers will increasingly but gradually look at their infrastructure costs in terms of opex versus capex, which will change the way they budget and plan as well as how they communicate with the investment community. This can be quite a change for some network operators who expect to purchase networking equipment once and then squeeze every bit of value that they can get from it over the next five to ten years or more.