The move to NFV has been underway for several years, and 2017 promises to see this accelerate even further. While service providers are fundamentally unique, there are some guiding principles that can be taken from any discussion of migration to NFV.
Deciding on which project or application to virtualize and in what order will depend on the risk and cost constraints of each provider. Here's a look at some of the issues and approaches service providers face as they decide why and how to embark on their journey to NFV.
The case for NFV is widely accepted, though sometimes misunderstood. A reduction in capital expenditures was long thought to be the largest incentive to pursuing network evolution. The initial premise was that merchant silicon and x86 hardware from white box OEM manufacturers would save tremendously on capital expense. It quickly became apparent this was not the case and that on a cost per bit basis traditional network hardware was cheaper. In the long run, however, the rise of capital expenditures may appear better in a trade-off of operational expenditures as the investments in physical infrastructure can support more customers and services.
Service providers' largest incentive may now come from the possibility of realizing massive operating expense reductions through adoption of new cloud-based and software-defined networking technologies. Service providers now stand on the precipice of achieving the largest leap in service provider networking technology in history. Cloud technologies were founded on the promise of achieving massive scale at extremely reduced costs. By repurposing these technologies into the network, not only do service providers get the ability to elastically scale the capacity on demand but also, achieve new scale and reduce current manpower footprint and inefficient legacy processes.
Many service providers have seen declines in average revenue per user (ARPU) for several years now. Reduced ARPU has led them to run their networks "hotter" by over-utilizing capacity beyond previous standards. At some point, something will have to give, as it will become untenable to sustain this level of service. One of the promises of SDN and NFV is to drive expanded profit margins (and fixed costs) through a reduction in operational expenditures as well as vastly increasing network service agility and speed to realization. This allows service providers to expand their offerings and create new ways to monetize new services.
Now that the case for NFV is established, the next step is understanding the criteria by which to select a project within the network to be migrated to NFV. In our work with service providers, we see two dominant themes emerging:
- Risk based: Most service providers have a conservative risk and rapid delivery approach toward research and development of SDN and NFV technologies. These consistently appear to be the dominant players in their market with a traditional or regulated competitive environment. In these cases, the service provider has enough capital or large enough projects to fund usually physical "greenfield," but virtual "brownfield" of the legacy network. Make no mistake that these service providers have no interest in virtual "greenfield" migrations as well. Many of them have several SDN/NFV projects going in parallel with some exploring different timelines for network evolution.
- Value based: We see a different research and development strategy in value-market or small-market service providers. They tend to select projects with the clearest business case, usually a very large and complex facility within the network. This presents an elevated risk of the delivered project being overdue and over budget. This is still logical, and once the project is completed, even if overdue and over budget,
The uncertainty of what will happen once a service provider embarks into the fog of NFV migration will never fully be cleared by discussion and research. However, making slow forays into the fog and in parallel driving the vendors to coalesce around fewer technologies, we may still navigate the rocky shoals of network innovation. In practice this is usually seen by small rapid evolutions of real-world work in trying to realize an NFV project. We see a number of service providers like Orange
taking steps forward.
Once a few inches of that fog has been cleared by accumulated knowledge, we can begin to make course corrections. We will continue to make more small rapid movements, and over time, break through that fog into a network that surpasses the reduced cost and service agility we originally thought possible.
— Mallik Tatipamula, Vice President of Service Provider Solutions, F5 Networks