The ability of companies like Google and Facebook to innovate and launch services quickly has long been coveted by telecom service providers. One key reason is the efficiency these companies have achieved in scalable data center operations by running applications on commodity and virtualized IT servers and storage. As a result, software-defined networking (SDN), which started in the data center, has been making its way outwards. Now, with network functions virtualization (NFV) — essentially software on commodity servers — operators can launch, deploy and scale services more efficiently and easily.
In effect, services can be provided in a manner that resembles cloud computing services with infrastructure being purchased and provisioned similar to Pods or “containerized systems” used in cloud data centers with elastic or tiered services. Beyond the highly anticipated savings of time, cost, and effort associated with today’s physical deployment of new hardware for new services, there is another major change made possible by virtualization: the way the network will be measured and priced.
While some operators have not yet come to grips with this aspect of elastic on-demand consumption, it is a certainty that service pricing will undergo fundamental changes. The traditional subscription model will likely be supplemented by a wide variety of per-occasion charges (premium and discount), and short-and long-term relationships that will come and go on a continual basis. The traditional long-term commitment subscription model may become extinct. Operators may also not have exclusive ownership of services, which can effectively be defined by any party with the right to do so, including a potential host of third-party developers.
For decades, networks and services have been constructed by linking a collection of purpose-built hardware appliances that work through communications trunks, driving cooperative behavior among network elements using proprietary software management systems. While the process worked (though not as well as it should) for traditional services, it is dilapidating in the age of the cloud and the Internet of Things. It can take more than 12 months for an operator to launch and scale a service, due to the complexity associated with coordinating the resource requirements, as well as integrating and deploying hardware appliances and software across the network. Often operators have to overprovision the network to ensure extra capacity is in place to support a service.
SDN enables operators to program the network by separating the system that decides where traffic is sent (the control plane) from the underlying systems that forward traffic (the data plane). With NFV, network functions are abstracted to enable them to run on commodity servers, switches and storage with all the benefits associated with that approach: cost reduction, flexibility and superior management. This decoupling provides operators the programmability, automation and control that have been sorely lacking to build highly scalable, flexible networks that can support new revenue-generating services to compete or partner with over-the-top (OTT) players.
It’s been just 19 short months since a group of carriers first published a document in October 2012 outlining the intent to move away from function- and vendor-proprietary gear to virtualized functions on commodity hardware. In that time, the progress has been remarkable, which is particularly surprising for an industry not accustomed to moving fast. Operators and vendors have been identifying which networking functions to virtualize on servers and the impact NFV will have on the network. The European Telecommunications Standards Institute (ETSI), the standards body behind NFV, is busy flushing out the architectural layers of NFV and the operational use cases. Experimentation of NFV is already taking place in 2014 and will lead to pockets of deployment in 2015.
The types of services enabled by NFV are by their very nature going to be multi-party and will include a number of physical and logical services (another reason why a silo infrastructure and silo OSS/BSS does not work). Chunks of technology in an NFV-driven network will have transient lives in different roles. End users and their service providers will also have transient relationships that will last only as long as they are needed. Flexibility in hardware and business relationships will create a competitive market that will benefit customers greatly. Although service providers may have a tough transition to this new reality, embracing innovation over regulation is the way forward.
Change will be dramatic and empowering. Product managers and network managers will implement services by programming the network, following a testing and validation process not much different from that used for deploying software releases. This will lead to faster and less risky service development and implementation. Product managers liberated from the constraints of service silos will quickly take advantage of this power and flexibility to dream up all sorts of new network services and service mash-ups. With these developments, they will likely bring in non-network service components, and with these services will come a previously unimagined array of novel priceable events.
In a world with NFV, we cannot easily envision all of the possible business models of tomorrow. What is certain is that the ecosystem will change, and companies that do not adapt will get left behind. Embracing change, leading the innovation charge and shaping the market will require OSS/BSS platforms that are service- and business-model agnostic to transition to the new cloud way of doing business. The impact on traditional hardware and software suppliers will be disruptive and extremely uncomfortable with the erosion of traditional procurement models as a result of NFV and SDN.
As procurement models for communications service providers (CSPs) change to include paying network equipment providers on a per-use basis for network functions versus capital expenditure tied to under-utilized equipment, some vendors will not be able to adapt fast enough. With this as a backdrop, hardware and software vendors alike are trying to reshape NFV and SDN to suit them. While traditional vendors rush to show the world how they can support virtualized functions tuned to their way of doing business, it is unlikely that all vendors will survive the likely consolidation. We’re all Cloudinians now and the tidal wave cannot be stopped.