Thinking About SDN Means Re-Thinking The Network
Software Defined Network (SDN) has been discussed for years. In this series, Craig Mathias looks at what impact it will have on Wi-Fi and why it’s needed.
Networks – the plumbing that makes the modern world possible – have a long history of innovation that can, at times, make the entire topic seem complex to the point of being downright obtuse. But, at its core, so to speak, networking really is simple – data goes in at one place, and, subject to a few rules, comes out at another. And the essential devices that make this easy-to-understand functionality possible – access points, switches, and routers – are conceptually simple as well.
In fact, it’s convenient and not at all incorrect to think of these devices as special-purpose computers. At the core of each is a processor running code (along, of course, with radios, specialized high-performance hardware elements especially in higher-end products, and clever management tools, each as appropriate), which brings up a very interesting possibility – programmability.
We tend to think of network equipment as manageable but otherwise functionally static, but network hardware has in fact become “softer” over the years as a more abstract, policy-driven approach to network management and operations has become dominant. The drivers here are such requisites as quality of service, class of service, and other performance- and capacity-boosting traffic management techniques.
So, suppose we take network programmability to its logical conclusion, and we build network equipment that’s dynamically reprogrammable, just like any other computer. And that, in a nutshell, is the essence of our topic for today – welcome to the world of software-defined networking (SDN).
At its core (there’s that word again; my apologies) SDN retains the essential functions of networking as we’ve traditionally known them, but adds a degree of (re-)programmability that can enhance flexibility, performance, security, cost-effectiveness, and much more, depending, of course, upon organizational mission, specific applications, and vendor implementation.
We can think of SDN as a new network architecture designed to keep pace with the growing, demanding, mission-critical nature of today’s organizational networks.
While SDN products have been on the market for some time, there’s still a good deal of controversy as to exactly what qualifies as SDN. Many network aficionados insist that support for the OpenFlow protocol is the very definition of SDN, but extended versions of OpenFlow and proprietary protocols are also on the market.
Some refer to SDN protocols as separating the control and data planes of a given network architecture – a concept quite familiar in the wireless-LAN world. It’s better, though to think of OpenFlow and other SDN protocols as enabling the specification of policy via a management interface that is supported across equipment from multiple vendors.
Regardless, it’s unlikely that OpenFlow or any other protocol suite will limit innovation in SDN overall – the concept is really that important.
A far more important question is how SDN will become the dominant strategy in an area of technology with deep roots and a long history of mix-and-match interoperability.
As we noted above, the drivers are likely to be a dramatic expansion in demand resulting from:
- BYOD (more users, and more devices per user)
- The requirement to facilitate applications with time-bounded traffic requirements (voice and video)
- The need for unified wired/wireless management
- The ever-present requirement for cost reduction.
We’ll explore the possibilities here in more detail in our next two columns on this topic, beginning with a really key question: just how will wireless LANs and SDN come together?
All Posts In This Series: