While the impact of software-defined networking (SDN) inside the data center is still reverberating among enterprises, SDN’s potential outside those walls is coming into clearer focus. The effect of extending SDN to the optical domain and beyond the data center could be tremendous in terms of enabling whole new applications and even business models for carriers and cloud providers.
With OpenFlow or some other SDN protocol establishing an open interface for unified control of the information technology (IT) and network resources both within and beyond the data center, individual slices of the network could be assigned to disparate parties for developing their own on-demand services, cloud-computing services, interconnection services for geographically dispersed data centers and other applications. Furthermore, such a scenario could unleash a whole new wave of networking innovation, with the development of custom networking applications that are not yet even envisioned.
What might extending SDN to the optical domain mean for carriers, cloud provides and large-scale enterprise network managers in the future? And what is required in terms of protocol innovation and connectivity assurance in order for that vision to come to fruition?
SDN’s Limitations and Possibilities
SDN is a concept for managing and virtually controlling networks, and it offers network operators terrific benefits in terms of streamlining and automating network infrastructure and operations and establishing a flexible environment for cloud-based networking and other services.However, SDN’s impact in terms of simplified network management and control to this point has been limited to the confines of the data center.
In most contemporary SDN deployments, open interfaces are integrated among centralized network management systems and electrical packet network elements. Control of the network elements is pulled out of those network elements themselves and into a centralized software device. A high degree of management and control software resides in the data-center infrastructure of packet switches and routers, and this allows SDN-based virtualization of servers and storage. Such forms of virtualization are being deployed across more types of enterprises every day, in fact.
OpenFlow is a simple, lightweight application programming interface (API) that is an increasingly popular protocol for such SDN implementations. With OpenFlow, flow-forwarding behavior is specified across an infrastructure of packet switches via an external controller.
The end vision of SDN proponents and the most radical new efficiencies that they envision, however, depend on extending these concepts and capabilities beyond the data center—to the circuit-switched, optical transport network. Today, for example, cloud computing is isolated from the control and operation of the optical transport networks that typically interconnect data centers. Cloud-computing processes and the statically configured network are not interoperable and do not interact. For network operators, this tees up a non-integrated, multilayer operation requiring individual, proprietary management solutions and tremendous costs with regard to scaling. It should be no surprise, then, that extending SDN to the optical layer has become a point of development focus for the optical-networking industry.
Bridging to the Optical Domain
A prototype implementation within the European Commission’s “OpenFlow in Europe—Linking Infrastructure and Applications” (OFELIA) collaborative project has provided a glimpse into a networking future in which SDN is extended to the optical wavelength-switched domain. The OFELIA prototype demonstrates integrated control of the network’s circuit-based optical and packet-based switching layers, all via a common, OpenFlow-based umbrella integrated with reconfigurable optical add/drop multiplexer (ROADM (News - Alert)).
The prototype implementation relies on additions to OpenFlow that adapt the protocol to the strict switching constraints of optical networking (wavelength continuity, optical impairments and optical power leveling on the line side, among them). The additions successfully allow OpenFlow to serve as a bridge between the optical and the enterprise application layers, effectively transforming SDN is into “SDON”—software-defined optical networking—and allowing true network virtualization across multiple layers.
For carriers, cloud providers and large-scale enterprise network managers, SDON based on OpenFlow or some other protocol might one day allow end-to-end, single-stroke orchestration of IT and network resources. The holistic control of all cloud computing, storage and networking resources could be enveloped under one, unified API, dramatically simplifying configuration, management and scaling of infrastructures. Introducing SDN beyond the data center would set up a scenario in which there would be less software content in the cloud or data center itself. Content could be located anywhere, with code accessing and running everything at the network layer.
With a constraint-aware agile optical network acting as a large, geographically dispersed switch, new opportunities stand to present themselves at the network control layer for large-scale, macroscopic network applications that haven’t even yet been envisioned. Bandwidth, latency and power consumption could be optimized per the needs of a given application, enabling true network virtualization and, with it, valuable capabilities such as capacity on demand, adaptive infrastructure and dynamic service automation.
Before that vision of a truly transparent, wavelength-switched optical network comes to pass in real-world commercial deployments, however, SDON requires additional development.
There remain issues, for example, with regard to the security and scalability of OpenFlow that will need to be addressed. The Open Networking Foundation (ONF) leads development and standardization of SDN and OpenFlow, and part of the ONF’s work now is concentrated on readying the protocol for the optical domain.
There also must be greater confidence in the performance of the connectivity beyond the data center. Customers will not move their high-value applications into the cloud unless they are offered the assurance of service-level agreements (SLAs) with regard to their connectivity to the cloud. SLA-based Ethernet into the cloud will be another requirement for network operators to maximize the business possibilities presented by SDON.
Furthermore, the prototype capability of OpenFlow-based SDON is available via the OFELIA testbed network to any researcher in the world for additional experimentation via standardized, secure interfaces through the GEANT network. Lessons learned there will inform ongoing development of the capability.
But SDON’s potential for carriers, cloud providers and large-scale enterprise network managers is clear: Extending SDN beyond the data center to the optical transport layer could play a key role in delivering the dramatically more efficient and scalable network operations that they seek.