The Promise of SDN

Is software-defined networking (SDN) a solution looking for a problem to solve? Or is it the next revolution in the way that networks are designed, operated, and managed?

The answer depends on whom you ask and when you ask it. On one side of the debate are such thought leaders of the neophyte industry as Stanford University’s Martin Casado, the "father" of SDN. They are using SDN to solve complex networking issues. On the other side of the debate are networking-hardware incumbents like Dave Ward, CTO of engineering and chief architect at Cisco Systems. They see both the potential and the threat of SDN and are integrating SDN features into their existing portfolios of products and applications.

The opportunity offered by SDN is that by separating the logic of the control plane from the forwarding in the data plane, it will allow faster innovation as the networking hardware and software are allowed to evolve independently. Additionally, this decoupling allows a centralized high-level program to monitor and control the network’s behavior in a holistic fashion.

However, these opportunities come with challenges, such as scalability in control plane elements responsible for a vast number of routing and switching devices, as well as reliability as control plane elements fail or become compromised.

It is said that the full potential of compute virtualization cannot be realized until the network has the same flexibility and programmability required to support complex distributed on-demand applications. Instead of today’s monolithic vertically integrated hardware- and software-networking elements with a distributed topology database, SDN proposes a different networking model with horizontally integrated networking elements and a centralized topology database. The transition from vertically integrated hardware and software to horizontal integration of the computer industry in the 1980s led to the PC revolution. Now SDN can enable the same causality in the networking industry.

One of the most prominent SDN solutions to date revolves around network virtualization. In this case, the physical network is abstracted into a logical IP backplane where virtual networks are programmed and overlaid like virtual servers in the compute space. This particular implementation of SDN is solving a number of network operational issues, primarily in the large enterprise data center domain.

SDN is without question a new networking paradigm with the potential to change the fundamental laws of networking physics. At the end of the day, SDN will be what we, the networking industry, make of it. We at Cox look forward to exploring SDN solutions in our network as business drivers dictate.

— Bill Coward, Principal Architect, Cox Communications

chrisd5110 9/5/2013 | 11:20:14 AM
Network Virtualization Bill,

You make a good point about SDN as an enabler for network virtualization.  In our research, we are finding that the combination of SDN and NfV brings significant value primarily in three areas:

* service agility - the ability to develop and provision new services quickly

* flexibility in the use of equipment and circuits

* operational efficiencies through common APIs and automation

I think you're right that this is a significant paradigm shift, and will be transformative in the years to come.
chrisd5110 9/5/2013 | 10:51:21 AM
Re: Pretty scary We've gone through similar transitions before, such as from circuit-switched voice to VoIP, or from physical servers to cloud computing.  It will take time for the technology to mature, but given historical precedents, I think these challenges will be addressed.
albreznick 8/29/2013 | 12:52:31 PM
Re: timetable Thanks, Bill. Any sense of where your cable brethren stand on SDN? I know there'll be plenty of discussion about it at SCTE in Atlanta in 2 months. Is this a 2014 or 2015 priority for the industry?
billcowardxx 8/29/2013 | 12:02:52 PM
Re: Pretty scary Yes, ping I agree, a radical change in how we approach networking today, It can be argued that our networks are reliable and stable but not very flexible when it comes to threats, bandwidth foresight, outages and end user programability. SDN must delivery the same network reliability and stability while increasing flexibility, agility and velocity.
billcowardxx 8/29/2013 | 11:40:59 AM
Re: timetable Cox is in the expolratory phase of enabling SDN technologies in our network and applications that will create value to our customers,  I'm not aware of a specific SDN timetable, but stay tuned.
sam masud 8/28/2013 | 1:13:15 PM
Pretty scary In the transition from a vertically integrated solution to SDN, we would, in a sense, being going from a "solid state" network to an eco-system with many "moving parts"--e.g. control plane, orchestration, apps, and also the addition of NFV elements. I would think reliability and performance of such a network would be and should be a top concern.
albreznick 8/28/2013 | 11:56:51 AM
Re: timetable More questions for you, Bill. When do you see the cable indiustry in general adopting SDN? How will cable ops use it? How will it work with CCAP? 
mendyk 8/27/2013 | 2:29:32 PM
timetable Bill -- What is Cox's timetable for deploying SDN? What stage are you at now?