Software-defined networking (SDN) and network function virtualisation (NFV) represent the most significant shift in communications networking in the past 50 years. Virtualised networking and software-controlled networking (SCN) can play a key role in the digital economy by helping operators reduce costs, increase responsiveness and create more innovative services. However, using virtualised networking and SCN to improve network operations requires a strong understanding of business drivers and the benefits of proper implementation. It also requires a realistic view of the factors likely to impede the rapid deployment of the new technologies and a sense of the likely path forward.

A recent research white paper by Analysys Mason titled “Software-Controlled Networking: Understanding the Implications of NFV and SDN for Operators” looks at the various business models that communications service providers (CSPs) must consider in their move towards adopting SDN and NFV while keeping in mind the various impediments and challenges…

Opportunities and time frames

Analysys Mason estimates that the SCN market (which includes SDN and NFV among other things) software and services market will be worth $13.1 billion by 2018. Some of the key opportunities presented by this market and the time frames in which they can be realised are as follows:

• Cost optimisation: SCN offers lower unit costs through improved infrastructural flexibility and power density while simplifying operations, as has been proven with cloud models. The research paper estimates that low-power, high-performance NFV and SDN hardware should become mainstream after 2016.
• Network systems and technology: NFV elements can be used to support new service priorities and drive businesses. These elements, starting in the mobile core, are likely to mature within three years and will be driven by long term evolution roll-outs. SDN spending will continue to focus on data centres through 2018.
• Service agility: SCN can facilitate the faster creation and launch of new services while decreasing revenue realisation times. While these are the stated goals for operators, investments over the next one to three years are likely to focus on cost optimisation.
• Revenue from innovations: Cloud-based on-demand delivery models suggest customer interest in on-demand network services. But the technology to provide such services economically and at scale has been lacking. By 2016, virtual networking and SCN should mature to the point that new service innovations will drive spends on cloud and SDN solutions.

The three main professional service opportunities associated with NFV and SDN are the services required for transformation or migration to an NFV and SDN environment, the management of a complex combination of SDN and traditional networks (including legacy), and the integration of different multi-vendor technologies for operations support systems (OSS) and business support systems (BSS).

NFV and SDN will require new and different carrier-grade hardware that is analogous to high-performance servers. SDN-enabled hardware will require more than the addition of an OpenFlow interface to an established box. Meanwhile, software, rather than hardware, will present the highest revenue opportunities for functions like virtualisation (network operating systems and hypervisors), OSS (SDN controllers, orchestrators and managers), NFV components (such as the policy and charging rules function [PCRF]),integration adapters and security.

Key risks in migration

Despite strong business drivers, Analysys Mason estimates that it will take three to four years to overcome all the inhibitors in SCN deployment. The main risks facing SCN adoption are as follows:

• Product readiness: NFV and SDN products have only recently been launched. It will take time to develop their features and capabilities, pricing and procurement strategies, deployment scenarios, methods and procedures of OSS/BSS; and to verify that service uptime requirements can be upheld.
• Lack of an assured business case: Opex reduction and new revenue are likely to drive the business case for SCNs. However, companies will require strong evidence before they commit funds, particularly because the total expenses could rise during the transition to SCNs. The transition from legacy networks to SCNs is likely to take a decade or more, during which physical and virtual domains must coexist and interact and be managed as a hybrid.
• CSP expectations of returns on non-depreciated assets: New technology initiatives generally have 10-year investment recovery cycles, and SDN/NFV elements are unlikely to replace assets before they are fully depreciated. Their replacement with NFV- and SDN-capable assets is likely to accelerate from 2018 onwards.
• Risks associated with fully open source solutions: The industry is debating the use of open source software for NFV and SDN implementation in CSP networks due to its likely impact on security and network management and the required standards. CSPs are trying to encourage open-source software through initiatives like the Open Platform for NFV Projects. It has taken three to four years for open-source solutions for data centre SDNs to be adopted, and it could take at least that long for NFV/SDN adoption.

Migration path

As new products come into the market, Analysys Mason expects that CSP spending will change more significantly in 5-10 years rather than over the next five years.

NFV spending will grow more rapidly than spending on SDN and cloud computing, primarily because SDN spending will be led by data centre providers (DCPs). DCPs will use more mature products and require less service investments than CSPs. Meanwhile, new virtualised OSS architectures will lead to an increase in CSP spends on NFV as they transform their networks to increase service agility.

At present, CSP spends on cloud computing are quite mature and not expected to increase at a higher rate than IT data centre spends worldwide. DCPs are seeking commercial off-the-shelf SDN hardware and software solutions that do not require the purchase of vendor services. Almost all SDN solutions are software overlay solutions over established SDN-enabled hardware. DCPs recognise that this adds an overhead cost layer, which they do not prefer. However, CSPs are accustomed to using overlay solutions to ensure that services are not disrupted.

The study expects the SDN revenue for CSPs to grow at the rate of enterprise SDN revenue through 2018 even though the solutions have been lagging behind by two to three years. Meanwhile, revenue generating SDN WAN solutions, such as cloud bursting and zero-touch dynamic VPN services, will drive CSPs and suppliers to increase overlay SDN deployments in transport (distribution) networks within the next five years. It has been forecasted that enterprise SDN spending by DCPs and CSPs will continue to be greater than CSP SDN spending even after 10 years (in 2023), given the size of the installed network base.

Milestones for OSS migration