SD-WAN: Abstraction Over the Wide Area

Arthur Cole
Fingers touching sphere

The enterprise is quickly learning what software-defined infrastructure can do for the data center. Now that the network can be provisioned as an abstract architecture along with compute and storage, organizations can finally rid themselves of the limitations that hardware places on data productivity.

Few enterprises, however, rely solely on their own data center anymore. Advanced cloud environments are placing increased pressure on wide area networks (WANs) to become more flexible and scalable – the very things that software-defined networking brings to the local environment.

The obvious solution, of course, is the software-defined WAN (SD-WAN), which extends the abstract environments across hybrid clouds and multi-site distributed data centers.

According to research house Kable Global ICT, more than 58 percent of enterprises throughout the world plan to adopt SD-WAN technology within the next two years. A key driver is the need to establish more versatile connectivity to branch offices and other remote sites, but it also has the potential to provide the kind of abstract distributed architecture that can meet the demands of everything from high-speed transactional processing to big data analytics and the IoT. With data loads not only becoming more burdensome but prone to greater fluctuations as well, SD-WAN technology also solves the twin challenges of dynamically scaling bandwidth consumption and maintaining connectivity to an increasingly ill-defined edge.

One of the key advantages that SD-WAN brings over traditional WAN optimization technologies is the ability to maintain connectivity across a broad range of network appliances. Once the data ecosystem pushes beyond the local data center, the enterprise no longer has complete control over infrastructure, so the network abstraction layer must be able to traverse an extremely diverse set of hardware platforms in order to provide the level of service that users require. This is equally true for the managed service provider and the carrier provider, both of whom must support near-universal connectivity in order to maintain seamless data operations for their enterprise clients.

The SD-WAN may also bring about the end of the network router, load balancer, WAN optimizer and other devices that lead to network complexity, according to technology consultant Steve Garson. In the emerging data ecosystem, organizations must support everything from traditional productivity applications to high-speed voice communications, each of which requires unique network characteristics regarding speed, reliability, signal deterioration and a host of other factors. Today’s network devices fail to take this into consideration, however – basically treating all packets equally. With a single SD-WAN controller, organizations will be able to introduce a high degree of programmability into network architectures that is driven by deep visibility into the data being transmitted. This will allow for the creation of customized network configurations on the fly using advanced automation and orchestration tools, providing each data workflow with an optimum network environment according to its needs.

The difference between a traditional WAN and an SD-WAN is like the difference between a map and an intelligent navigation platform, says Taneo Consulting CTO Marc Sollers. One provides a view of how to get from here to there, while the other collects data on traffic congestion, road construction, weather patterns and a wealth of other points to select the optimal route. In this new era of dynamic network configuration, organizations not only gain the most efficient use of resources for network traffic in the present, but information as to how infrastructure, architectures and even the data itself can be altered to continuously improve performance going forward.

The true potential for the SD-WAN can only be surmised at the moment. As field experience increases, it is likely that the technology will grow and evolve in ways that not even its designers could have anticipated – such is the flexibility of software.

In the emerging digital services economy, success will usually go to the most nimble, and that can only come about when the last links to static physical infrastructure are severed across the entire enterprise data footprint.

Related articles