Service Provider SDN Gets Real
While software-defined networking (SDN) is filling column inches as the industry rushes to promote its benefits, very little is heard from operators with concrete plans for real implementations of the approach. This is grist to the mill of SDN skeptics, who believe that SDN belongs in the realms of theory and won't be applied to existing service provider networks anytime soon.
But pioneering communications service providers (CSPs) are determined to prove the skeptics wrong. They want to understand sooner rather than later how SDN concepts perform in the real world and whether the approach can deliver on its promises. And they plan to turn market leadership on SDN implementation into competitive advantage.
Such CSPs are also reading the writing on the wall with respect to their network operational costs and investment. These are rising faster than revenues in many cases, and the gap is widening as traditional communication services revenue declines and demand for capacity to support third-party data services (from which CSPs derive little or no revenue) soars.
The pressures are mounting, particularly for mobile operators, as Ericsson's latest Mobility Report (June 2013) shows. According to the report, mobile data growth -- which doubled between 1Q12 and 1Q13, is set to double again this year, and Ericsson forecasts a compound annual growth rate of 50 percent for mobile data between 2012 and 2018. Meanwhile, mobile voice traffic -- for so long the engine room of telco revenues and profitability -- is only growing at 4 percent.
The traditional way of building and operating networks hugely delays the development and launch of new services, preventing CSPs from differentiating themselves quickly in the market and failing fast when necessary. Nor can they easily adjust their networks to specific customer and application needs. As a result, third-party service owners use the network inefficiently, and CSPs lose out on revenue-generating opportunities to enhance their customers' network experience.
Forward-looking CSPs are taking the bet that SDN will help them to reduce costs and create new, network-based services. They hope that by decoupling network control functions from the forwarding plane, service innovation will no longer be constrained by the slower development pace of networking hardware. They anticipate that by exposing network capabilities through APIs, they will encourage an increasingly rich ecosystem of third-party application developers to build compelling apps. And they like the idea of slicing the network into virtual segments that match the needs of specific applications, which will help them differentiate traffic flows in innovative, cost-effective and monetizable ways.
It's clear that advanced CSPs take a holistic view of their networks -- they want to apply SDN concepts across network layers and domains, so that they can program it in flexible ways, end-to-end. Ericsson has created a tagline for this vision -- "service provider SDN" -- to differentiate it from SDN initiatives that focus on virtualizing the data center network.
CSPs that are fast-tracking the adoption of SDN within their networks and are willing to talk openly about what they are doing and why, are doing the industry a huge service. SDN is exciting, as vendors such as Ericsson suggest, but it is also daunting for those CSPs trying to implement it. There is an urgent need for knowledge about what works and what doesn't, how SDN will coexist with the legacy network and -- as is usual at this early stage of a major market disruption -- how best to make the business case.
Telstra Corp.'s stance is helpful in this respect. It is not only a pioneer, but a vocal proponent of SDN. The company isn't working alone: It has been collaborating with Ericsson on its first SDN use case -- service chaining -- since 2012 and acknowledges the support that a large vendor with longstanding credentials in the SDN domain can give. It also has a clear timescale for introducing SDN technology into its network. Telstra expects to move to field trials later in 2013 and to take the first steps toward operationalizing SDN in its production network in 2014.
Telstra is trialing a service chaining capability that uses Openflow to support the dynamic sequencing of network services. The SDN theory Telstra is testing is whether Openflow-based service chaining can support application delivery more cost-effectively and with a higher level of customer experience than current approaches to configuring the network. Telstra wants to create a flexible and dynamic network environment in which packets can be routed through specific -- and reusable -- network services on-demand.
Service chaining use cases typically mention firewalls, DPI devices, virus scanning and content filtering as possible service examples. Among the benefits of this approach, Telstra anticipates reduced cost of service delivery, more surgical scaling of network services to match demand and an ability to offer richer services that are tailored to customer requirements, Customers will be able to self-select the components of these services. Each service component will represent an opportunity to add value to -- and collect revenue from -- customers.
Though Telstra fully expects SDN implementation to be a challenge, it is confident in its ability to manage this. Change is a constant in the telecom business, and many operators have become highly practiced at handling it. SDN merely represents the next stage in the continuous evolution of their infrastructure.
Ericsson points out that service chaining is one of a number of SDN use cases a CSP might choose to implement, depending on their business priorities. CSPs with growing cloud businesses they want to support more efficiently, for example, will concentrate first on SDN in the data center network. Other CSPs wish to virtualize specific cost centers within the network or areas that are initially easier to address than others, such as the home gateway or aggregation network. Whatever the starting point of an SDN project, partnership and knowledge sharing are likely to be important to its success.
Leadership and confidence will also be crucial. CSPs own immensely valuable assets that are critical to modern life. Yet the way those assets are architected and run are increasingly out of step with the rest of the world. Service provider SDN is a sensible proposition, but it needs leaders in both the operator and vendor communities to step up and make it work. The sooner the industry can learn from real SDN implementations, the faster it will start to close the revenue/cost gap threatening to overwhelm it.
— Caroline Chappell, Senior Analyst, Heavy Reading