Software-Defined WAN: Reality or Just More Hype?
With the current architecture in place for more than 30 years, the WAN certainly is ready for change.
I've been involved in networking for about 30 years now, having cut my teeth as an engineer during the Internet's rise, when the constant stream of innovation made things really exciting. But then networking matured, and the industry settled into a period of relatively little innovation -- let's face it, a jump in speed is about as exciting as listening to Republican presidential candidate Ben Carson.
Things started to get interesting again about five years ago, with the idea of the software-defined network (SDN) giving rise to a bevy of networking startups. SDNs are threatening the status quo, promising to bring unparalleled levels of network agility alignment with applications and business processes. SDNs are a panacea to all network woes -- or at least that's what we've been told.
For all the hype, SDN deployments are still few and far between. Networking professionals who I know even joke that SDN actually stands for "still done nothing."
Out of the Data Center
Why is the gap between SDN hype and reality so big? I believe it's because traditional SDNs (if such a thing exists) are focused on the data center where the stakes are high. Also, data center SDNs require businesses to assemble a wide variety of technology to get the solution to work. We'll get to widespread SDN data center deployment eventually, but the journey to that point is long and slow.
Over the past two years, SDNs have pivoted and become focused on the enterprise WAN. The SD-WAN brings the benefits of SDNs to the wide-area network. In the WAN, software-defined solutions have a chance of narrowing the gap between reality and hype. Most organizations struggle mightily with managing the WAN, and trends such as unified communications, cloud computing, and mobility simply add to the complexity of today's WANs. All companies have local engineers and technical staff in data center locations but the WAN connects branch offices that tend to have little or no staff.
Lastly, the WAN is ready for change as the current architecture has been in place for more than 30 years. Network engineers designed the existing model to handle that era's client/server traffic, which generates predictable, consistent levels of traffic. Chaotic systems, like the Internet, generated only a small portion of WAN traffic, as organizations kept almost all important systems on premises. The centralized architecture is optimized for manageability but lacks any kind of dynamic bandwidth allocation or granular access control required for today's diverse and varied traffic types driven from mobile users, cloud platforms, and collaborative applications.
The evolution to digital business is putting new demands on the WAN that legacy models cannot meet. WANs are currently expensive to run, use bandwidth inefficiently, have little to no automation capabilities and are not optimal for cloud traffic. Changing a network can be difficult for many companies, since the business is often the network, but digital organizations need agility. Not taking the step to create an agile network will hold such organizations back.
Think of a traditional WAN as a network that is optimized for compute-centric IT models, like client/server systems. An SD WAN is a next-generation architecture optimized for network-centric compute models. One of the biggest benefits of an SD WAN is that it enables application policies to be used to automate configuration updates, move traffic flows, or enact other changes to ensure the network continually meets business needs.
However, as is the case with all emerging technologies, the topic of SD-WAN has become a confusing one for many network managers. Almost every vendor that has a stake in the WAN now claims to be an SD-WAN solution provider. The term "SD-WAN" has taken on multiple definitions (we in the communications space should be used to that since UC went through a similar phase). And SD-WAN deployment options vary.
Some businesses will be aggressive and jump into SD-WAN with both feet, either via premises-based infrastructure or with managed service. Others may choose to run in a hybrid mode, mixing and matching traditional and software-defining networking, for years.
I will be moderating a panel on SD-WAN at Enterprise Connect 2016, taking place next week in Orlando, Fla. The following smart people will be joining the discussion: Travis Ewert, SVP of network software development at Level3; Kevin Gavin, CMO of Talari Networks; and Mike Wood, VP of marketing at VeloCloud. If you're like most companies and struggling with managing your WAN and you're looking to learn more about SD WAN, please come and attend this informative session, "SD-WAN: Networking for the Cloud-based, Mobile, Real-time Enterprise. Thank you and see you in Orlando!