Transport SDN: It’s Time to Get Real

In recent years, much has been debated about SDN, and specifically transport SDN and its applicability in modern networks. Let me provide you my perspective on this topic.

It is well known that the separation of the network control plane from the data plane brought by SDN enables faster introduction of innovation and vendor-agnostic, unified, and dynamic network control.

SDN-based control increases service velocity and provides a greater ability to control service levels, therefore improving customer satisfaction, accelerating revenue, and potentially enabling new revenue streams.

A multi-vendor, multi-domain, multi-layer management and control approach enables optimized resource usage and more efficient network operation, allowing for OpEx and CapEx reductions. This methodology has been applied successfully for some time by cloud providers for setting up connectivity within data centers and in data center interconnection. But what about transport networks?

Packet optical transport networks include a photonic transmission layer where equipment specific techniques are often used to maximize optical channel performance and optimize spectrum use. Current common data models do not provide enough optical equipment modeling detail to allow for the ultimate level of network optimization using an arbitrary “generic” SDN controller.

Additionally, the design of older transport networks, as well as the limited reconfigurability of some legacy optical equipment, limits the level of programmability of the network.

You will find more infographics at Statista

Is transport SDN a dead end, then? I am not convinced it is, and here is why: an estimated 4.5 billion people use the internet today, and expect instantaneous, on-demand access to a variety of cloud-based applications.

This generates enormous amounts of data that is stored and processed in large data centers that are distributed around the globe. If we add the Internet of Things data generated by sensors everywhere, and the fact that enterprises are moving their workloads to the cloud, it is easy to see how connectivity between data centers, as well as connectivity between data centers and end users, is evolving fast. This evolution is as much about increasing demand for capacity as it is about the ability to cope with rapidly changing traffic patterns.

With 5G, this trend will only be reinforced: the definition of different service types, some with strict service level requirements in terms of latency or reliability, relies on dynamic real-time assignment of hardware resources to the new demands (network slicing), down to the optical layer.

The latest generation of optical equipment, including coherent interfaces, is also stepping up, offering new levels of programmability and enabling an intelligent software layer that not only assigns resources to services in real time but also finds their optimal operation configurations.

However, the steps taken today when an operator plans and activates new services are far from providing the level of flexibility and agility described above. This is mostly addressed by over-dimensioning and provisioning ahead of needs.

But in a scenario of increased network complexity and continued decline of revenue per bit, operators need to use the network’s resources as efficiently as possible, while managing operating expenses. Operators need help from transport SDN.

And here is the good news: despite all the challenges mentioned above, transport SDN is already helping operators today. Infinera’s Transcend transport SDN solutions have been deployed in a wide range of field installations, providing visible benefits to network operators around the globe. I have recently presented a webinar on Real World Transport SDN Success Cases, covering some interesting ways transport SDN is being used today to answer the needs of Infinera’s customers. These include the following:

Operators need: Unified network topology and visualization, equipment, and inventory of services

Infinera delivers: Transcend software collects and reconciles network information originating from different equipment technologies, suppliers, locations and NMSs. The discovery and consolidation of all network data under a single database, combined with powerful graphical tools, offers real-time end-to-end network and service visualization and generation of thorough network reports.

Operators need: Support for planning the migration of a legacy network to modern infrastructure

Infinera delivers: the consolidation of network information under Transcend mentioned in the use case above enables network-wide analysis of multi-vendor legacy networks, as well as the evaluation of network migration needs and the definition of priorities. When the analysis is complete, the migration is planned with the help of the Transcend workflow engine, according to the requirements of different service types or geographical locations. The migration can then be executed with the support of the Transcend Maestro Orchestrator. Orchestrated migration is a suitable solution to ease the complexity of network migration activities and reduce the process’s risk.

Operators need: End-to-end path computation and service provisioning in multi-vendor, multi-layer scenarios

Infinera delivers: Transcend implements a hierarchical multi-vendor, multi-layer, multi-domain path computation engine (PCE). The hierarchical PCE combines information from various network layers, establishing the resource relationships required to find feasible routes across layers between two endpoints. The optimal path is selected by minimizing a user-configurable abstract cost factor, while conforming to user-defined constraints. Once the optimal path is found, Transcend provisions the corresponding service from end to end, providing efficient resource use and simplified operation.

Operators need: To resolve or minimize the dropping of packets during peaks of resource congestion

Infinera delivers: Transcend Adapt closed-loop automation engine is used to perform continuous dynamic optimization of network resource use and balance the link load. The routing of new demands uses real-time link load information to avoid the most congested links and, under certain circumstances, some services are rerouted through less loaded links. The result is a balanced use of network resources that is enough not only to eliminate packet drops but also to enable up to 25% additional network throughput without any added hardware.

Operators need: Increased network resiliency to multiple failures without new/dedicated hardware

Infinera delivers: Transcend vASON implements best-effort dynamic service restoration, with shared resource use, at Layers 0/1 or 2.5. The SDN-based service restoration makes use of Transcend’s customizable PCE for full control of path selection, while supporting configurable restoration triggers.

It can be used to provide service restoration upon failure but may alternatively be used to compute backup routes ahead of the next failure, for faster protection-like traffic switching. The resiliency against multiple sequential failures in meshed networks is significantly increased, without resorting to dedicated hardware installed and statically reserved ahead of time as would be the case with classic protection schemes.

These are some examples of state-of-the art transport SDN deployments that have already brought benefits to operators. When looking for their common aspects, the following key ingredients for success can be identified:

• A focus on concrete network problems or operator business needs and the adoption of a stepwise, incremental approach to automation.
• Close cooperation between the vendor and the operator, as the former knows the solutions best while the latter knows its network best.
• Support by a knowledgeable and experienced professional services organization that can customize a solution and ensure integration with existing back-office systems and network equipment.
• Modular software that supports open, standard APIs and makes use of common data models, for seamless integration into operational environments and automation ecosystems that will grow with the operator’s needs.

I think it is fair to say that by evolving these individual use cases and making the most of the latest hardware and software evolution paradigms, transport SDN will hold its promise to enable a fully programmable network and create an increasingly more automated infrastructure.

And now that you have read about these inspiring automation success cases and their key ingredients, it’s time to find out what Transport SDN can do for your network!