Communications service providers (CSPs) are aggressively driving Network functions virtualization (NFV) into their networks and are optimistic about its value and success. In contrast, many CSPs are skeptical about software-defined networking (SDN) and its role in their networks. Here are some thoughts from James Feger of CenturyLink, who takes a pragmatic view of SDN.
In the past few months Feger has made a number of public statements about SDN at conferences like TIA, where he famously said, “The reality is that we’re several years away from an environment where SDN saves the world as everybody is proclaiming.” This was widely viewed as a pessimistic take on SDN, but I see it as a more pragmatic and realistic view. CenturyLink is actively engaged in activities related to SDN (and NFV), so Feger is speaking from the viewpoint of experience.
I recently had a chat with Feger to get some more insight into his thinking. He noted that as a service provider with customer-serving production networks, CenturyLink cannot simply “go out and replace existing equipment to gain SDN capabilities.” This requirement to support the current infrastructure really limits the options, and Feger noted that they “are just now starting to hear about strategies for supporting and integrating legacy equipment.” Even if they could start from scratch, Feger noted that there is limited availability of SDN-enabled equipment. As equipment evolves, there will be opportunities to deploy SDN capabilities. While SDN features alone may not justify replacement, they are certainly desirable for future applications.
We talked a bit about where in the network SDN can bring the most value. Today, much of the talk is about propagating SDN from the data center to the network backbone or core. However, Feger notes that there are not a lot of activation and configuration changes at the core of the network. In contrast, the edge of the network is where services originate, and that is where SDN may enable some very valuable automation and service innovation.
One good example of how SDN might facilitate service innovation is in dynamic private cloud services. Today, CenturyLink offers private line connections to cloud services in its own Savvis data centers, as well as in third-party hosting facilities. Feger noted that these connections are very static today, and are achieved using traditional network-to-network interfaces (NNIs). He added that SDN offers the possibility of more dynamic private (VPN) connections to cloud services, which could be ordered on a self-serve basis just like today’s cloud services. With SDN, today’s static interconnect could be replaced with a dynamic connection controlled by the customer’s order. This is the type of innovation that makes SDN interesting.
Finally, what about OpenFlow? Feger believes that “SDN is much broader than OpenFlow” and that some within CenturyLink “are not sure about the state of scale/support/adoption that OpenFlow is going to have in the long term.” Personally, Feger does have some skepticism about OpenFlow, saying that “OpenFlow is a little bit of reinventing the wheel. Do we really need to have another protocol? We have an opportunity to look at the variety of existing signaling mechanisms before we just wholesale discount their capability.”
I believe that James Feger and CenturyLink are taking a very aggressive stance on trying to modernize their network and to accelerate their creation of services. Even so, they are realistic about how quickly this can be done and when the benefits could be realized.