The color of SDN: Photonic dimensions to software-defined networking intelligence

As carriers look beyond such initial implementations to the practical application of software-defined networking (SDN) principles in the wide area, and more specifically end-to-end, optical-based service infrastructure, a new set of challenges – and opportunities – come to light.

Aug 29th, 2013

Industry interest in software-defined networking (SDN) shows little sign of abating. This is not surprising when one considers the market forces that underpin interest in the technology and its promise.

Hyperscale growth – driven by mobile broadband, HD/OTT video, and cloud computing – is creating an unprecedented level of disruption in carrier networks. This disruption stems not only from the massive volumes of data, but even more importantly from the increasingly unpredictable and dynamic nature of the bandwidth-hungry applications that drive those volumes. And these applications come, more and more, with stringent service-level requirements.

In this evolving network operations environment, simply providing more capacity between stress points in the service infrastructure is no longer enough. Requirements can change much more rapidly than before, with significant fluctuation in demand between any points in the network at any time. Traffic engineering in multi-layer and multi-vendor networks therefore has become more operationally complex, especially in the optical domain, where physical constraints and service planning requirements are very different than in packet-based networks. This operational complexity slows provisioning times, increases operational costs, and ultimately affects a carrier’s ability to seize the revenue opportunity that hyperscale growth represents.

Initial interest in SDN has centered on specific use cases, in particular high-capacity data center networks. Here, data volumes and application characteristics such as unpredictable traffic patterns and workloads provide fertile ground for SDN’s promise of increased efficiency, flexibility, and scalability through centralized service management, network virtualization, and resource abstraction. In these packet-transport-based scenarios, SDN provides the framework for simplified service management via Layer 2/3 functionality (e.g., centralized routing, path calculation, policy management) and multi-vendor interoperability via standards-based protocols (e.g., OpenFlow). Using open APIs, SDN decouples the network control and forwarding functions to enable network control to be directly programmable and the underlying infrastructure to be abstracted for applications and network services.

As carriers look beyond such initial implementations to the practical application of SDN principles in the wide area, and more specifically end-to-end, optical-based service infrastructure, a new set of challenges – and opportunities – come to light.

Laying the coherent foundation
The advent of coherent optical transmission has set the stage for an entirely new optical service infrastructure. Coherent 100G, an ideal granularity that is well proven in DWDM networks today, leads the way in achieving higher data rates per optical channel to cope with traffic growth and network scalability. The technology provides longer reach, lower latency (critical for data center interconnect), easy upgrade paths, and a simplified network layout by reducing the number of client interfaces needed. It raises spectral efficiency within the network as well, creating much-needed extra capacity. Coherent 100G also leads to significant operational and capital cost benefits for operators through simplified operations, reduced energy cost per bit, and lower site costs. Coherent 100G is also backed by a fully developed ecosystem, including standardization, components, and compatibility with the ITU wavelength grid. Despite coherent 100G being a relatively new technology, the industry is already looking ahead to the requirements and solutions for 200G, 400G, and even 1-Tbps coherent transmission.

As evidenced by this progress, the physics behind optical transmission has, and always will, play a critical role in the evolution of optical networks. From an operational perspective, the more important question is how can operators harness this pool of coherent optical capacity to better serve demanding new transport applications and meet end-users’ heightened expectations for more dynamic, real-time bandwidth services? And how do they do this without increasing network costs and complexity?

Beyond SDN: Maximizing end-to-end service value with optical layer extensions
As adoption of coherent 100G becomes more widespread, SDN intelligence at the transport layer will play a critical role in enabling operators to efficiently harness “agile” service capacity, improve resource utilization (helping to reduce capex), and create new revenue opportunities by simply using end-to-end flows within the optical domain more intelligently and cost-effectively.

To achieve this, two very important extensions to software-defined intelligence in the transport domain need to be addressed. These extensions provide the foundation for unlocking the full potential of an agile, software-defined optical network, while maintaining the resiliency and availability requirements of traditional TDM-based transport networks.

Horizontal extensions: Tapping into fully automated end-to-end workflows and operation processes
As operators know well, the physical constraints and challenges inherent in service planning at the optical layer are mirrored by the complexities of operations and workflow management. For example, the introduction of a new wavelength service might call for complex physical network planning, with additional interface cards that make it necessary to deploy personnel to each site. Keeping the status of equipment and planned services synchronized across all the necessary management “silos,” such as field personnel, planning, inventory, and network management systems, is a major process. This often results in DWDM networks that are not only rigid and static in physical terms, but also rigid and constrained in the operational sense. The result is poor utilization and stranded capacity.

Today, SDN proponents tend to focus heavily on service creation. But in optical networks, the whole process – from physical commissioning all the way to maintenance – needs to be addressed to maximize the performance of the network and yield the full benefits of software-defined intelligence and automation. This includes real-time interaction between planning and network management. Enhancements to the SDN framework that enable SDN controllers to fully capitalize on automated end-to-end workflows (Layer 0 to Layer 2.5) and procedures can not only simplify optical layer service creation, but also improve efficiencies in network planning, installation, and management.

A coordinated, real-time, comprehensive (i.e., “global”) view of optical layer operations and process management (including spares handling) will enable network operators to:

  • reduce overall operations costs
  • provide optimal use of deployed systems (resulting in reduced capex)
  • optimize resource utilization (e.g., wavelength, reach, slot space, 3R pool, etc.)
  • accelerate service provisioning cycles, which ultimately leads to much faster time-to-revenue.

These unique dimensions in software-defined intelligence and automation at the optical layer have not yet been fully considered within the SDN framework. Yet they are critical to efficient end-to-end service architectures.

Vertical extensions: Maximizing optical innovations in the SDN framework
Recent advances in transport technologies have brought a new level of flexibility, intelligence, and control of service capacity to the optical layer. In addition to coherent optical transmission, these advances include flexible ROADMS, flexible-rate transponders, and flex-grid and superchannels. Colorless, directionless, and contentionless ROADMs (CDC ROADMs), for example, enable any wavelength to be switched in any direction and independently of all other channels. Meanwhile, flex-grid and superchannel technologies enable operators to dynamically adapt the wavelength grid to the needs of high-capacity, long-reach transport superchannels (e.g., via channel bundling), as well as increase spectral efficiency, and thus the capacity of transport systems.

In addition, with enhanced integration of multi-layer switching technologies (OTN, MPLS-TP, SONET/SDH, and Ethernet), network resources in the optical service infrastructure are better utilized and more efficiently integrated with the photonic layer than ever before. By extending the SDN framework to enable a tighter coupling between an intelligent and automated photonic layer and higher network layers (including the application layers), operators can create multi-layer service architectures optimized for flexible and dynamic access to optical capacity – at different granularities and with customized SLAs.

Physical layer automation and intelligence provides the basis for this coupling, and extensions to the SDN framework will enable network operators to easily extract and more fully use optical layer capabilities and innovations. They will also enable real-time interaction between a dynamic Layer 0 and the business applications requesting access to a more agile, on-demand pool of service capacity.

SDN and intelligent optical control
Coherent optical transmission is the new cornerstone for high-capacity, high-performance optical networks, enabling long reach and low-latency transmission. Despite advances in photonic layer capacity, reach, and system performance, however, most optical networks today continue to lack the end-to-end flexibility and efficiency required to meet the needs of new and demanding applications such on-demand data center interconnect.

In addition, hyperscale growth continues to burden operations staff with increased operational complexity. This growth further compounds multi-layer service delivery challenges and significantly impairs an operator’s ability to create and sustain competitive differentiation.

To fully optimize multi-layer architectures for the flexible provisioning of high-capacity services, network operators need to take a more comprehensive approach to software-defined intelligence at the transport layer. Such an approach should include service-centric extensions to the SDN framework that harness optical layer flexibility and simplify service planning and management processes such as workflow automation.

Uwe Fischer is chief technology officer at Coriant.

More in High-Speed Networks