This week I attended the Linley Tech Carrier Conference in San Jose. There were many interesting presentations on trends in Cloud RAN and HetNets, along with details of new basestation SoCs and network processors. Given 6WIND’s focus on solutions for Software-Defined Networking (SDN), I was especially interested in a presentation by Christos Kolias, Research Scientist at Orange, on “Service Providers in the Era of Software-Defined Networking”.


Kolias started his presentation by outlining some of the key trends for service providers worldwide. He explained that, while CAPEX is declining for wireline services, it is increasing for wireless, with overall wholesale revenue margins shrinking as a result. Mobile data traffic has now surpassed mobile voice traffic, while sales of smartphones were double the sales of PCs in 2011, projected to grow to 1.5 billion units in 2016.


Given these trends, Kolias explained that service providers face major challenges in terms of traffic (more traffic does not mean more revenue), the sheer number of overlays in Telco networks (MPLS, ATM, VPV, IMS, SONET, CDN etc.), the OTT providers (Skype, Google Voice, YouTube, Hulu, Netflix, Facebook etc.), the opportunity to provide value-added services (managed hosting, security, conferencing, cloud services etc.) and the overwhelming need to move to a business model that is ARPU-driven rather than subscriber-driven. An ARPU-driven business model would be based on bytes rather than minutes and could also leverage the wealth of subscriber data that is available within the service providers. Ideally, bandwidth would be sold as a utility, like electricity. These are all valid concepts but in general the industry does not have strategies to achieve them.


In terms of SDN, Kolias presented this as a new paradigm and a shift in networking that allows for true innovation. The ability to remotely control network hardware, using software, leads to real programmable networks, while the decoupling of the control and data planes creates an abstraction and logical map of the overall topology. The use of standard (i.e. vendor-independent) interfaces to the hardware has the potential to change the way that service providers architect and build networks.


So what are the business drivers for SDN, from the perspective of a service provider? Kolias focused on cost (increasing APRU while minimizing both CAPEX and OPEX), performance (in terms of customer service quality, uptime, automation and resiliency), security, compliance and the ability to innovate and grow.


Kolias mentioned three interesting use cases for SDN. The first was traffic offloading: unpredictable traffic spikes are a serious network problem and SDN enables traffic flows to be managed intelligently based on criteria set (and modified dynamically) by operators with a macro-level view of the network.


Second, mobility management enables seamless roaming across heterogeneous wireless networks. Kolias showed a fascinating video of a car driving around Stanford with uninterrupted video streaming due to GPS-assisted handovers between WiFi and WiMAX, as well as flow rerouting and replication.


Finally, Kolias discussed the use of virtual networks (or network slicing). This concept enables the network administrator to set up and manage virtual networks where, for example, one would be dedicated to IPTV, one to VoIP, one to DSL and one to mobile traffic.


In summary, Kolias stressed that SDN provides major benefits to service providers through bringing intelligence, programmability and flexibility into their networks, while noting that SDN architectures will co-exist and complement the existing infrastructure through a hybrid approach (hybrid switches and hybrid networks). There are no plans to “rip and replace existing boxes” any time soon.


Were you at the Linley Tech Carrier Conference this week? What discussions did you find most interesting? Which of the many new product announcements will be most significant for carrier networks?