Ask most enterprise executives if they are interested in deploying SDN, and you'll probably hear a solid “yes.” Ask them what they plan to do with it, however, and the response will become a bit less clear.
Most will highlight network efficiency and flexibility, lower costs, the use of white box infrastructure and a range of other benefits, but a specific vision of exactly how SDN will support applications and business models remains very much a work in progress. And in the end, it isn’t entirely clear whether SDN will improve today’s IT environment or sweep it away altogether.
One of the key areas that SDN is expected to influence is dev/ops. By bringing networking up to the virtual, abstract layer, organizations will be able to create and scale test environments much more quickly and cheaply than before, pushing new products and services to market in record time. But at the same time, as InfoWorld’s Serdar Yegulalp learned at Delphix’s “State of DevOps” conference last month, the entire notion of what dev/ops is and how it functions is under scrutiny. Development teams are starting to encounter real-time streaming, live data and other factors, partially as a result of SDN, and fracturization within the field is making it difficult to tell exactly what kind of network architecture is best for each project.
Others see SDN more as a play toward greater resilience and reliability in distributed data environments. Sonus Networks’ David Tipping sees a wide range of business continuity benefits resulting from abstract wide-area networking. These benefits derive primarily from the creation of redundant pathways without the expense of provisioning additional MPLS bandwidth. SDN can also improve centralization and automation of things like network orchestration, governance, visibility and traffic management. This should ensure quality end-user experiences without over-provisioning the network.
Meanwhile, SDN development is moving towards giving applications greater control over network configurations, reducing the need for managers to manually tailor architectures for a given function. The Open Networking Foundation recently announced two “intent-based” northbound interface (NBI) projects, dubbed Aspen and Boulder, that would allow apps to specify their needs without having to carry network information wherever they go. The app conveys its intent to the SDN controller, which delivers the proper configuration from a pool of open resources. ONF officials say a key application will be to support QoS for real-time multimedia and Unified Communications services.
But while it is tempting to view SDN in terms of solving today’s problems, there is a strong likelihood that it will also foster entirely new applications, architectures and even business models, says ACG Research analyst Michael Kennedy. For one thing, and as difficult as it may be to accept, SDN will produce the most dramatic gains in efficiency and cost by reducing the human factor in data networking. At the same time, removal of expensive ISRs and other devices will save thousands and ultimately alter the very workflows and business processes that drive much of the activity in the digital economy. This will be highly disruptive but is necessary if current enterprises are to compete with startups that are being built for abstract networking from the ground up.
As with any technological development, then, SDN must be viewed in terms of its impact on workflows and processes rather than technology. If the case cannot be made as to how a particular deployment will improve results or lower costs, then it is probably wiser to wait until a clear rationale has emerged.
But think hard, and don’t wait too long. Ready or not, the change to abstract networking is just around the corner.
Photo courtesy of Shutterstock.