Software networks are universally recognized among network operators as a necessary investment to be technically ready and positioned for future 5G networks and, consequently, they are progressively embracing the shift to virtualization. In 2016, mainly thanks to both the maturity of open source projects and the initial deployments ran by major service providers, they accelerated their investments on the softwarisation of their networks. During the last two years, the market has seen trials, proofs of concept (PoC) and demonstrations showing some early and small-scale success stories. There are currently significant efforts underway across the major communication service providers worldwide, such as Telefonica, British Telecom, AT&T, China Mobile and Deutsche Telekom, to roll out new 5G infrastructure based on NFV/SDN technologies.
In relation to commercial solutions, there is an initial generation of them put forth by telecommunications vendors and supporting technology partners who are trying to expand their products by adapting to the disruptive change in the traditional telecommunications value chain. The majority of the main incumbent solution vendors (such as Cisco, Ericsson, Huawei, NEC, Nokia, Oracle, etc.) have created NFV/SDN platforms and virtual network functions (VNFs). There even are new insurgents that arrived from the IT world, looking for the new market opportunities offered by these new technologies and the virtualisation of telecom infrastructures.
The scene has continued showing multiple open source NFV projects too. Some of them, like Open Platform for NFV (OPNFV), a project led by the Linux Foundation, have a broad aim of defining the NFV ecosystem; others have a more focused approach. The development of MANO systems has been one of the toughest challenges faced by the virtualization community and, as a result, a number of industry projects and programs have emerged during the last years: Open Source MANO (OSM) hosted by ETSI; AT&T's ECOMP (Enhanced Control, Orchestration, Management and Policy) and Open-O (Open Orchestrator Project), that merged into one project under the Linux Foundation, namely, the Open Network Automation Platform (ONAP).
Special attention must be also paid to quite a few projects funded under European Union R&D programs, among which SONATA is included.
As it can be easily observed, there is a variety of NFV orchestration approaches competing for market leadership, including offerings from leading service providers, large network equipment providers, IT and systems integration firms and open source communities.
In relation to standards, there are some groups that have done significant work to push standards closer to completion. NFV standardisation has been driven mostly by ETSI NFV ISG, while ITU-T pushes for SDN standardization and IETF/IRTF community is active in both areas. However, these standardisation bodies are not the only organisations working on the development of open standards for NFV and SDN. On the contrary, as we have commented previously, there is a large number of other SDOs, industry consortiums, open source initiatives and research projects involved in developing standards and creating guidelines for SDN and NFV.
Despite all this dizzy activity, that the market is evolving, and that technical expertise is starting to mature, reading about NFV/SDN seems to be an exercise in contradictions these days. Some articles claim the explosive progress in SDN and NFV deployments while others talk about a deceleration in adoption and even about the loss of confidence in the technology by telecom operators. The truth is that the NFV/SDN adoption is being relatively slow and that the technology rollout is still in early stages today.
Much has been written about the benefits that NFV/SDN technology will bring to telecommunication networks, but there are also some issues and concerns that may have led to the delay in adopting these technologies and may continue to slow down the wide adoption. It seems that the hype/disillusion/reality progression model of the Gartner hype circle is a general fact also applicable to software networks.
For a monolithic sector like telecoms industry, with huge technological baggage and legacy infrastructure to consider, virtualization is not only a big challenge but a cultural shock as well. For example, the adoption of the DevOps model needs to create cross-functional teams in communication service providers. There are also challenges in linking virtual networks to existing operations and business support systems (OSS/BSS). These barriers, together with the nature of telecom networks, considered critical infrastructures, the hyper-regulated telco environment, and the cost reduction crusades in many operators, may have been the main reasons for partial deployments rather than full adoption and commitment from many operators. This halfway approach does not allow the service providers to get the full benefits of virtualization, further slowing down the deployment following a vicious circle.
The lack of detailed NFV standards, especially in the critical management and orchestration (MANO) stack, is another thing to be considered. Despite the work done so far, it is a reality that comprehensive standards for NFV/SDN are not available yet. The number of organisations working on NFV standards and orchestration is not helping operators either. All these initiatives, claiming their own merits and wonders, are driven by well-respected organizations, communities, vendors and tier-one operators, some of which are even participating in a few of them simultaneously. Service providers are unsure of which NFV solutions are most likely to gain acceptance and, thus, be worthy of the considerable investment and risk associated with choosing an NFV orchestration platform. In addition, open source initiatives do not use standards but specifications that are not even complete yet. This makes the certification process a challenge, which means that the desired interoperability is not possible yet, what becomes even more complicated if we take into account that NFV challenges current telco certification processes. For more widespread adoption to occur, there is great need to open, interoperable standards that will help service providers future-proof their deployment roadmaps.
Additionally, communication service providers traditionally rely on vendors for their deployments but, understandably, they want to avoid vendor lock-in and they are becoming more pragmatic in their plans, with a greater insistence upon vendors embracing standards-driven design approaches and also the adoption of cloud-native practices, such as containers, micro-services, etc. Although open source is gaining higher acceptance among communication service providers, they typically take the “just buy and use” approach. Hence, they would not usually choose an open source alternative if they couldn´t outsource the support of it. In particular, smaller CSPs don’t have the personnel to engage directly and contribute to the communities, so they will probably prefer to work through their vendors, who will have a more direct role with the communities. Furthermore, open source solutions usually require additional development/integration work for adoption. But vendors face the same dilemma by having to choose which initiatives to back and what role to play. Network and IT suppliers have to make sure that their NFV/SDN products are well positioned in this disruptive new market and they often backup multiple open source initiatives. The move towards open source standardization adds another level of complexity and unease for them. Instead of focusing on proprietary products, there needs to be a stronger emphasis on creating new services but also on integrating and customizing other third-party solutions, for example.
In addition to the above-mentioned barriers, we need also to highlight that many of the new desired/promised features and innovations, such as slicing, recursiveness, scalability, etc. are on their way but not really delivered yet.
This market structure makes it challenging for service providers to pick the winning long-term solution. While a small number of operators are pioneering the rollout of software and virtualization technologies, many are postponing their investments waiting for technologies to mature, and for evidence of the actual benefits before making any commitments, which is also affecting vendors.