Network Function Virtualization — NFV — promises much. NFV is the next generation of network architecture, replacing software running on dedicated hardware with virtual appliances running on commodity hardware. NFV should mean the ability to deploy software functions on any hardware at any time — at the touch of a button or the turn of a dial, and with high interoperability so any number of functions can be switched and mixed between vendors. But this promise is not being realized.
The classic telecoms network model is a collection of dedicated hardware — multiple chunks of metal that all need individual installation and maintenance, per box, per site. This is the old way of building a network — heavy on capital expenditure, long on planning cycles, and usually ending in vendor lock-in and uncompetitive pricing due to the complexity of interconnection. Upgrades and replacements are also resource-heavy in terms of integration and commissioning.
So it makes business sense for the network world to move towards the NFV model; an approach that provides rapid and agile installation, and an evolution that will help deliver predictable operational spend and ultimately enable communication service providers (CSPs) to benefit from the associated reduction in the total cost of ownership.
Software interoperability
To really make this work though, the software elements need to be fully interoperable, in order to enable vendor independence and competitive pricing. The resulting network is rapidly scalable, flexible, and benefits from dynamic resource allocation. This is what NFV should be enabling – the access to a full range of interchangeable best-of-breed, trusted Virtual Network Functions (VNFs) that can be easily and cost-effectively deployed.
What is actually happening is that a lack of information and insight means that CSPs are becoming locked into full stack virtualized solutions from a limited set of vendors. Instead of having their choice of hardware constrained by lack of interoperability, they are now finding constraints in the virtual world as their choice of software is being stifled through lack of accessible, certified information.
NFV interoperability is nowhere near the level it should be, but there are ongoing attempts to address the problem. NFV MANO (management and orchestration) is the architectural framework that should facilitate the deployment and connection of services as they move from dedicated physical equipment to virtual solutions. However, the ETSI descriptors are only partial, enabling vendors to maintain some proprietary elements. This means there are gaps in the information used for virtual function operation.
Similarly, the ONAP project aims to create a comprehensive platform for real-time, policy-driven orchestration and automation of physical and virtual network functions. Its goal is to enable software, network, IT and cloud providers and developers to rapidly automate new services and support complete lifecycle management. However, the influence of the major vendors is very powerful, and they are still steering the industry to create a minimum set of elements so they can maintain opacity for proprietary elements.
This means that sourcing from more than one vendor becomes problematic. Compatibility is not a given, and integration that should take minutes is instead taking days, or even months, because the onboarding of each new virtual function is needlessly complicated by this partial interoperability. And vendors are changing software releases all the time, so every time a new release is issued, the painful onboarding process has to take place all over again.
We also need to remember that this move to virtual networks is not a revolution, but an evolution. Of course, there will be greenfield NFV-only networks, but there are also billions of physical network elements out there that still have extensive useful lives. NFV can help these brownfield networks evolve with market requirements by putting virtual functions in place where flexible capability is needed – capability that can be turned on or off as required. This move to hybrid networks that combine physical assets with virtual functions will bring enhanced operational efficiencies and speed-to-market for new kinds of services.
Today, the reality is that a customer buying VNFs can spend months trying to integrate the functions, and the uptake and progression of NFV is suffering as a result. NFV should be unleashing the power of innovation, and this will become increasingly important as we move closer to 5G networks and virtual functions that push computing power to the edge. For applications such as augmented and virtual reality, serious edge computing power is needed — 5G cannot do half of what it promises to achieve without NFV.
NFV know-how
What the industry needs is an authoritative source of information on NFV solutions, with a consistent, descriptive searchable taxonomy. In addition, there needs to be a service that tracks and administers certification and standards compliance, so compatibility can be determined before implementation. Without a definitive registry, NFV will not have the transformational effect on our networks that it has the potential to deliver.
By introducing a single point of reference to compare VNF policy and software versions, network operations will be improved, and network security will be enhanced with the ability to identify broken, malicious or outdated VNFs. Network operators will be able to better understand in advance which functions will work in their specific environment and can be integrated quickly to catalyze the benefits of NFV.
Confusion and frustration has delayed service provider deployments of NFV. The complexities of connecting legacy physical components and initiating virtual methods for provisioning, ordering, assuring, maintaining and billing services are not being clarified. Lack of information, lack of automation, and too much vendor lock-in are limiting the value proposition of NFV for communication service providers.
Gartner says that the promise of NFV will drive ‘robust revenue from digital services and enhancements.’ If that is to happen in the future, then it’s time for industry to standardize the information and the tools required to create, manage and operate virtual and physical network assets and functions effectively. It’s time to centralize the expertise and to streamline and simplify the NFV landscape. The result should be reduced complexity, elegant scalability, the clean automation of business processes and the profitable optimization of assets.  Only then will NFV become the unstoppable shift that will catapult operational efficiency and increase customer satisfaction.
