Software-defined networking (SDN) is on the tip of everyone's tongue, and new initiatives seem to roll out daily from the networking vendors. The concept introduces strikingly different ideas for the networking discipline and is hailed by some as a huge innovation. But underneath the hype, there are practical use cases SDN is suited for and problems it could solve. Getting there, however, will take a little time.
Whenever I talk to someone about SDN, the word "revolutionary" inevitably comes up. The conversation sometimes begins sounding absurdly like the lyrics from the classic Beatles song:
You say you want a revolution
Well, you know
We all want to change the world
Don't get me wrong. I believe that SDN is a disruptive technology that eventually will transform our current idea of networking. But, like the Fab Four, I'm not cheering for an imminent revolution.
Instead, I'm anticipating a long-term migration to SDN that will occur as organizations update their infrastructures and, more importantly, discover how it can help them practically manage network resources. At least that was my thinking before Cisco recently announced its Application Centric Infrastructure (ACI). With ACI, SDN seems more achievable in the near term. But for most businesses, the transition will be an evolution, rather than a revolution that takes place overnight.
As the technology is implemented, we will see computer networks that are more dynamic and easier to manage. That's because SDN allows you to abstract your underlying network architecture. You can turn a network -- or at least its SDN-ready components -- into a service for providing connectivity and compute resources on an as-needed basis. There is also evidence SDN will make load balancing, virtualization, and cloud deployments much simpler, and that high levels of utilization will be baked into the infrastructure. SDN will also provide much greater visibility into the network and associated servers, storage, and applications.
But realizing these benefits will take time because SDN is still a nascent technology. Some networking prognosticators are saying we need one to two years for practical products to be accepted by the market. And that sounds about right to me.
Having outlined those qualifiers, I don't think it's too early to start planning for how you can take advantage of SDN. There are some interesting ideas out there for using SDN to increase the flexibility and capabilities of infrastructure-as-a-service (IaaS), virtual computing and storage, and datacenter load balancing.
Using SDN in the real world
To prepare, you need to start creating use cases to educate your team and change how they think about the network. A couple good examples of SDN scenarios come to mind.
The first involves the healthcare industry, which is under intense pressure to improve care and lower costs. Hospitals and other provider organizations are investing heavily in technology for applications such as big data, mobile diagnostics, electronic records, robotics, and telemedicine. All these strategies are compute, storage, and bandwidth intensive.
In today's world, IT managers for hospitals and other large institutions struggle to sustain performance during periods of peak usage. But with SDN, it is possible they could deploy fewer IT resources and still keep everybody happy. For example, you could program the network and your compute environment to shift resources to patient registration and hospital operating rooms in the morning when demand is typically high.
Later in the day, you could shift them to patient recovery and critical care, when those departments are busier and need more from the network. During the night, SDN could redistribute IT resources for bill processing, updating patient records, and other activities. Essentially, SDN is performing load balancing and virtual management on the fly so that resources are fully or better utilized.
The second use case focuses on the wired classroom. One of the big challenges for schools is providing adequate infrastructure to support technology-enabled education. With SDN, network administrators could gain flexibility for dynamically managing resources at the class and individual-student level.
Such capabilities provide opportunities for sharing resources across campuses based on the needs of students and educators. SDN also lets you grant security on a micro level for individual students or classrooms to permit access to specific resources, files, or websites during tests or sanctioned learning activities.
Industry providers are looking at these types of use cases and seeing significant opportunity for creating new revenue streams. There is currently a lot of discussion about what type of hardware and software will dominate in SDN. Cisco is betting that it has the answer with ACI. The company also has a solid list of application ISVs, and, in some cases, even competitors are partnering to develop on its open platform.
It is also possible we will see solutions based on proprietary networking software running on commodity hardware. But other network providers won't bow out of this market quietly. Regardless of the market leaders, SDN will have a dramatic impact on enterprise networking, not because it's the latest trend, but because it is, well, revolutionary.