As they begin to transform mobile operator networks, network functions virtualization (NFV) architectures are typically touted as a means of offering significant reduction in capital and operating expenditures, and there’s little question that this is true. Reductions in these expenditures remain a chief and primary driver for moving away from proprietary hardware-driven networks. There is another huge motivator that’s only now beginning to show itself as a key driver behind operator enthusiasm for shifting to NFV: service velocity, and the resulting democratization of applications and services that come from a NFV architecture.
Prior to NFV, service providers such as mobile operators had an all-too-real incentive to minimize the number of mobile applications they deployed, because every new application required a new piece of hardware, and would have to be brought in separately, tested separately and implemented separately. This process is exceptionally difficult and complex to manage.
In the recent days of hardware-driven environments, it often wasn’t worth it for an operator to buy or deploy an application with an unpredictable number of users – for instance, a solution for the enterprise, a solution for MVNOs, or one targeted at a particular user sub-group. The fear was always that there might be stranded hardware with only a small number of users on it, and; therefore, a surefire money-loser. However, when an operator deploys virtual machines, they’re allowed to be much more custom and less risk-averse, enabling them to deploy specific applications they might not be able to justify if they’d otherwise have to buy new hardware and go through all the incremental effort and cost of bringing it online.
A cloud-driven, NFV-centric approach allows for a quick response to changes in demand for applications. Let’s say an operator launches a sports video application for soccer/football, a sport that happens to drive its peak traffic only around a couple of events per year or even every four years. Cloud-based NFV architectures let operators flexibly “burst” capacity to the cloud to ensure a high level of quality for users who only wish to watch the World Cup, the Champions League final and the Real Madrid/Barcelona match live on their devices, but who generally consume at otherwise normal levels during the rest of the racing year. Operators may find that they need to triple capacity overnight for these big events, and only a NFV architecture allows them to do this in a rapid and flexible manner.
This might also be where MVNOs and enterprise applications come in. An enterprise with a specific application that’s unique to them may want to run that application on the mobile network infrastructure. In the recent past, service providers were justifiably skeptical: “We only expect a few hundred thousand users – how is it in our best interest to bring this in?” With NFV, turning up a new application, even one with a traditionally-defined small user base, is an extremely straightforward and repeatable process. Furthermore, these targeted applications often deliver a far higher margin than mass-appeal applications that tend to become commoditized. As long as the service provider has a virtual machine available, they turn it up, they run it, they insert it into the call flow for that specific customer – and it works.
NFV also makes “service chaining” much more straightforward – in which a whole set of network services are delivered, and together make a new application possible. Each service is enabled and managed via settings in orchestration software that allocate how many resources are provided and to whom. Only the resources that are needed are allocated, and at times when they’re not needed, they’re simply distributed elsewhere – perhaps to some other Enterprise application that might need them.
It goes without saying – but I’ll say it anyway – that it’s generally much faster to make changes and enhancements in software than in hardware. In the “old days” of the all-too-recent past, operators were hamstrung by having to use vendors’ ASIC- and FPGA-based implementation of services they were interested in deploying to users, which have a long lead time for engineers to develop and fabricate.
By contrast, when using a NFV server-based platform those engineers and partners gain flexibility from instead working with software, which is easier to add, remove and evolve in order to put new functionality in place. This helps greatly in making sure that new, innovative and market-differentiating services can be delivered within needed timelines – and that a true democratization of services and applications can expand within operator networks worldwide.