Abstraction is kind of a big deal in software-defined networking (SDN). It is a useful tool in simplifying and effectively shielding the complexity of the network’s internal attributes to enable rapid service fulfillment.

However, for efficient, reliable, and secure network engineering operations abstraction is less useful, while transparency is highly valuable. The network operators and their support systems must discover the full nature of the transport infrastructure in order to analyze and best use it.

The real network topology must be discovered at all layers down to the physical. A multi-layer SDN controller can solve many issues in network engineering operations by providing this discovery function and leveraging the discovered information. From this knowledge, many efficiencies and optimization opportunities flow.

New service opportunities are emerging in the context of distributed cloud architectures, data center interconnect, and 5G networking, for example. All this will be underpinned by IP/optical transport and networking services which must have massive scale and the ability to be delivered rapidly, securely, reliably, and with the contracted QoS.

These and other evolving requirements of the IP/optical network are driving a renaissance in IP/optical integration. This renaissance is being seen principally in the control and management layers. This is enabled by a number of developments and advances in networking and management technology, but two in particular are very significant:

• The new generation of optical hardware, bringing major advances in automation and in instrumentation capabilities;
• The introduction of SDN bringing evolved, centralized, hierarchical control, and management software architectures.

These advances enable the discovery of the exact nature of the transport infrastructure. Many operator challenges can be tackled and mitigated by these technological transitions.

In IP/Optical networks discovering, capturing, and accurately visualizing the combined topology and interconnection points of the network, from layer 0 to 3, has been a significant challenge to operators. Organic growth over time — often driven by disparate departments — has led to complex, poorly understood network architectures. These bring the potential for severe operational issues such as unpredictable failure impacts on crucial network traffic and sub-optimal latency control.

Operators need reliable tools to discover, visualize, and explore multi-layer topologies and inter-connections, which will allow them to better use their network assets. This is vital information to allow the following challenges to be overcome:

• Inefficient traffic routing: Operators must gain greater and more dynamic control of their network assets so networks can be run more efficiently and services can make the best use of available bandwidth. The gap between service provisioning and network engineering must be bridged so that operators can satisfy dynamic service requests while making optimal use of available IP/optical assets.
• Lack of effective protection: In IP/optical networks, the full set of risk scenarios in the optical transport layers (Layer 0, Layer 1) must be discovered to simplify scenario planning and boost network reliability. A multi-layer protection strategy must be devised so that traffic is protected at the most cost-effective layer without the need for expensive 1:1 redundancy, for example.
• Cumbersome maintenance coordination: An automated multi-layer maintenance procedure is needed to minimize, or even prevent, negative impacts to user traffic during maintenance activities.
• Sub-optimal latency: Latency information must be gathered from the optical layer in order to aggregate the transport layer contribution to the overall delay budget to allow optimization to meet service level agreement (SLA) targets.

SDN is a transformative technology delivering real value particularly in applications requiring a broad span of vision and control. Within emerging SDN architectures multi-layer SDN controllers are appearing to optimize the operation of networks across optical and IP layers.

Many optical devices now use embedded operation, administration, and management (OAM) protocols to reliably detect and communicate details of topology and connectivity to the SDN controller. The Link Layer Discovery Protocol (LLDP) is used by network devices for advertising their identity, capabilities, and neighbors.

Known as LLDP ‘snooping’, this is a valuable discovery tool for data center interconnection applications and WAN architectures. LLDP snooping has many benefits including avoiding cumbersome manual configurations and verification of the configuration. It can also detect misconfigurations. Heuristic techniques, which can be less deterministic, can be used to discern the topology and connectivity of the network layers where LLDP is unavailable, for example in legacy installations.

These discovery capabilities can enable a solution set to address the challenges identified above.
A multi-layer SDN management architecture can bring a coherent and unified view to the management of IP/optical networks.

A centralized, global view enables clear insight into the way in which traffic is engineered onto network paths. This permits comprehensive route computation to take place with the goal of best using network resources for maximum efficiency.

A typical rendered architecture shows services overlaid on the IP and optical layers. Detailed information can be accessed from both the IP and the optical domains and fresh insights into network operation can be synthesized, particularly in the area of risk assessment. Path diversity may be apparent at the IP layer, but deeper insights into the actual optical infrastructure can reveal shared risks, for example where two apparently disjoint optical paths traverse a common fiber, duct or optical layer forwarding device.

This visualization can be used to ensure minimal impact to IP Services during planned maintenance activities. These activities at the IP and optical layers can be coordinated and orchestrated, for example, to keep IP traffic away from affected optical links during the actual maintenance.

The ability to discover and aggregate contributions to latency in the transport network brings real competitive advantage, particularly as we enter the 5G era. Many 5G use cases, such as AR/VR and remote system control, depend on consistent and low latency.

While SDN brings a range of tools to the automation and optimization of communications networks, multi-layer discovery within SDN can bring deep insights into network operations. Because what you don’t know can in fact hurt your network.