To thrive in the cloud era, network service providers (NSPs) must make their services as easy to consume as the cloud services that now dominate network traffic. They need to adopt a dynamic network service delivery model that allows customers access to what they want, when and where they need it. They need to innovate at cloud speeds, moving from concept to market in days or weeks, not years. And while everyone will agree that software-defined networking is the solution, simply making a decision to adopt SDN is just the start of the journey to dynamic network services.

Inevitably, the next question is: How? When it comes to extending SDN into metro- and wide-area networks (MANs and WANs), there are two choices: overlay or underlay.

The overlay model, which is the basis for many of today’s data center SDN products, is being driven by cloud and business services groups within NSPs, and by providers with many disparate networks. The proposition is simple: Leverage SDN solutions in the data center to extend dynamic creation of tunnels to other data centers and to remote sites across any intermediary MAN or WAN. The advantage is network transparency. Cloud providers can instantiate applications and IT infrastructure services that reach any customer, anywhere – without regard to who owns the underlying network infrastructure. NSPs that lack network reach can rapidly extend their enterprise network services to reach their customers, wherever they may be.

The underlay model, more commonly referred to as Carrier SDN, is at the other end of the spectrum. It is being driven by the network operations groups within NSPs that want the ability to fully control the network itself, with direct linkages to every layer from wavelength/OTN to MPLS/IP, and across all vendors. Carrier SDN will oblige them with two major capabilities.

First, it will provide network abstraction and standard data models to minimize the OSS-to-network integration complexity that is stifling service innovation and provisioning. A simplified OSS means NSPs can add support for new network elements and new network features in days or weeks, not months or years. Automated service provisioning can finally break through technology and product boundaries to encompass multi-layer, multi-vendor networks.

Second, it will provide global, real-time network visibility and control so network assets can be used more efficiently. Carrier SDN will assess the state of the entire multi-layer network before mapping service requests to transport links. This will ensure customer requirements for latency, bandwidth, etc., can be met, and optimal use is made of IP/optical network assets. Guided by operator-specified policies, path computation engines will be able to create new paths to meet spikes in demand, avoid potential failures, support new, on-demand, pay-as-you-go services, and make the most of unused bandwidth.

Will the two approaches square off against each other and instigate the industry’s next great SDN debate? Not likely. Each has a role to play in empowering NSPs to deliver new services and take control of their networks. And interworking between overlay and underlay SDN models is key to integrating new virtualized network functions with legacy networks. It’s not any perceived battle between the two but the realization of the potential they both bring to the table that will make this an exciting space to watch in the coming year.