As we look forward to 2016, I can confidently say it is time for NFV to move out of the lab and into the real world. It is technologically sound and mobile network operators (MNOs) know it makes sense from a business perspective.
Undoubtedly, NFV has long been a case of "when" rather than "if," and it is fair to say from the proof of concept (PoC) experiences that Red Hat has had with carriers in recent times, the answer to the question "when?" is "Now."
Operators are actively moving from a lab environment to pre-trial, depending on the maturity of the customer. It will come as no surprise to those who have been watching the space that the Evolved Packet Core (EPC) is seeing the greatest traction when it comes to NFV.
EPC is proving popular because mobile operators are in the process of making additional LTE deployments, so as they look for a second wave of infrastructure deployment to phase in a new technology, it is great timing and a good justification. With NFV proven in the labs, MNOs are looking at whether virtualization is an option in the field or whether they have to stick with traditional hardware.
Quite how NFV will be deployed remains to be seen. In all likelihood we'll see MNOs taking different approaches. It is a different economic model whether MNOs use a virtual network function (VNF) hosted in the cloud at the edge of the network with less hardware or whether they use fiber to link back to a regional data centre that hosts the software. Some will use one solution from one VNF vendor while others will use multiple solutions from multiple vendors. This freedom is only enabled by taking a scale-out, cloud-based approach to NFV.
After 30 years or more familiarity with network equipment provider (NEP) hardware practices, the understandable stumbling block to date for MNOs looking to make the move over to NFV is a people issue. Regarding the transformation from tried-and-tested traditional telco kit usage to an IT-centric approach, generally speaking virtualization expertise was in the IT ops center but not the network center. MNOs have to find the right people and train those people. Operators we see are at different maturities, so they have different blends of technologies and people. Even though carriers are working with IT companies, they still need to build in a lot of expertise that they did not have beforehand.
It's important to note here that while there are certain regulatory drivers regarding LTE rollouts, NFV implementation is not a race for operators. Whether you're a fast follower, early adopter or even late to the party, the most important thing is to ensure it is watertight and MNOs know this, hence the caution. We are working with operators who are close to production and others who have not really started.
NFV for the operators we're working with is not about reducing costs -- at least not in the short term. It is all about agility and new service introduction. Because NFV is not an immediate cost-cutter, it can be a difficult business decision to make. That said, there isn't a single MNO without an LTE strategy, and that means they all have NFV plans.
Today what we're seeing is operators building one NFV infrastructure (NFVi) platform but doing just one VNF or one service. This gives them the room to grow, although it could also be structured as a single VNF that enables multiple service components. NFV rollout will be done piece by piece: Realistically, we're going to see a hybrid model for many years to come. It's very likely that hybrid model could last forever, mirroring developments in enterprise computing, where vital legacy applications might be running on mainframes while more software-centric deployments will be in the cloud.
What is of paramount importance for MNOs moving forward with NFV is to avoid becoming locked into a vendor. It does not make sense to unlock yourself from a network equipment provider by taking advantage of NFV only to lock yourself onto a proprietary virtualization software stack.
It possibly goes without saying, but the carriers we're working with are taking an open approach. They have taken a strategic decision at the outset not to further invest in pure traditional hardware but to move to OpenStack.
Plenty has been written on the subject of OpenStack, including here on Light Reading, and the ability to deliver multivendor NFVi-to-VNF interoperability is critical to the success of a scale-out approach. And as the results of Light Reading’s 2015 NFVi-VNF tests demonstrate, the benefits of the collaborative approach taken by OpenStack participants have boosted the level of interoperability between the different vendors and their respective services. (See EXCLUSIVE! NFV Interop Evaluation Results.)
As an industry, we first started talking seriously about NFV in 2012 when ETSI published a white paper on the subject, promising "to outline the benefits, enablers and challenges for Network Functions Virtualisation (as distinct from Cloud/SDN) and the rationale for encouraging an international collaboration to accelerate development and deployment of interoperable solutions based on high volume industry standard servers."
In fairness to the industry, four years is not an especially long time to wait. This rapid speed to market has almost certainly come about as of result of NFV being universally approved of as being beneficial to the industry. Now that the PoCs are complete, field trials are likely to be rapid. The good news is that the wait is over -- we will see commercial deployments this year. The fact is, MNOs can't afford not to make a success of NFV.
— Sandro Mazziotta, director of NFV product management, Red Hat Inc. (NYSE: RHT)