A defining moment for SDN

As traditional, and even more modern, network architectures struggle to cope with more dynamic applications and services, another acronym has emerged as a possible answer to operators’ challenges. But ask ten different industry pundits what SDN (software defined networking) means and you will likely get ten different answers. At the time of writing, telecoms.com had just published a poll offering a variety of definitions, the clear leader of which was “other”—effectively meaning something else not on the list.

To some it’s a concept, a system architecture or even an implementation of technology, to others it merely means the OpenFlow protocol used to configure the forwarding plane of a network switch or router over the network. Then again, some see it as a process of virtualisation—the creation of an abstraction layer between software and hardware in the network.

This last point is one many can agree on. Telecoms networks typically contain a variety of proprietary hardware solutions that have grown in number over time, as new network services often require additional kit, and the integration and deployment of these appliances is getting harder to do.

Meanwhile hardware life-cycles are becoming shorter as innovation accelerates, reducing the return on investment of deploying new services and constraining innovation.

SDN and the virtualisation of network functions aims to address these problems by evolving standard technology to consolidate many network equipment types onto high volume servers, switches and storage—essentially running the virtual applications on commodity hardware.

This is the task standards body and industry specification group ETSI set out to undertake in January, when it proposed to develop an architecture for the virtualisation of various functions within telecoms networks.

The initiative is led by seven operators; AT&T, BT, Deutsche Telekom, Orange, Telecom Italia, Telefónica and Verizon; which have been joined by 52 other network operators, telecoms equipment vendors, IT vendors and technology providers to create the ETSI Industry Specification Group (ISG) for Network Functions Virtualization.

ETSI will focus on implementing network functions in software that can run on a range of industry standard server hardware, and that can be moved to, or instantiated in, various locations in the network as required, without the need to install new equipment.

Potential benefits include reduced operator CAPEX and OPEX; reduced time-to-market; improved return on investment from new services; greater flexibility to scale up, scale down or evolve services; openness to the virtual appliance market and pure software entrants; and opportunities to trial and deploy new services at lower risk. The first specifications are expected before the end of 2013. When we spoke to Ulf Ewaldsson, CTO of Ericsson, in October 2012, he said the explosive growth in cloud technology was spurring the requirement for SDN specifications—a view reinforced by a recent Ovum report. As the three-tier hierarchy (access, aggregation, and core) of network architecture is being replaced by flatter architectures, virtualised application software is replacing network appliances, and network infrastructure is becoming more ‘programmable’. Ovum believes that SDN provides an opportunity to completely re-examine network architectures, introduce virtualisation, and provide truly innovative solutions, with more of a focus on the intelligence inherent in the network, rather than the feeds and speeds of data.

With SDN the network will dynamically adapt to provide the connectivity services that best serve the application and a better approach will eventually produce networks that are much more flexible in providing new services and monetising the network, as well as being more efficient in their use of resources.

Yet hype is an ever present threat in the world of technology and Informa senior research analyst Dimitris Mavrakis expects some observers to say, “this feels like 2006 all over again.” The thing is that telco SDN essentially promises the same thing that IMS promised six years ago: horizontalisation of the network without clear revenue opportunities.

However, he states, “telco SDN has serious advantages over IMS: SDN is already being implemented in the IT domain and operators and vendors will have learned from their involvement in IMS. Also, SDN is attempting to enter the market by converting practices and technologies from the IT domain so that they can be applied in the telecoms environment. Contrary to this, IMS was a completely new— and very optimistic—concept.”

This won’t necessarily stop vendors creating fanciful and colourful presentations about SDN, nor stem the talk about how operators need to move from silos to horizontal platforms and networks.

Addressing the cacophony of different definitions for telco SDN is ETSI’s Network Functions Virtualization (NFV) forum. The NFV believes Network Functions Virtualisation is applicable to any data plane packet processing and control plane function in fixed and mobile network infrastructures, although NFV sees itself as complementary to Software Defined Networking in that the topics are mutually beneficial but are not dependent on each other. But although SDN and NFV are arriving at an appropriate and interesting time for carrier networks, as telco and IT environments are merging, prompting a requirement for clear definition, Mavrakis believes telco SDN currently has some serious challenges to overcome.

“Each vendor has a different definition of SDN, but hopefully NFV will force vendors to align to its vision. NFV compatible (or standardised) elements will not arrive in the market for at least one to two years from now. And NFV elements may require forklift upgrades for existing infrastructure, where there may not be a clear revenue opportunity, but only cost savings,” he says.