I admit it. I’ve grown weary of the debate about whether SDN includes network programmability or whether or not SDN can only be accomplished through NfV, or the relative merits of control plane / dataplane separation. I will leave those debates to others more focused on the technology itself. Personally, I have been more fascinated with what I see as the new business opportunities emerging around SDN.
Certainly there is a raft of opportunities for start-up companies in the controller space or in the virtualization of various networking functions. Many innovative new companies are re-examining existing network functions within the SDN paradigm; that will lead to some potentially new and useful approaches that may be cheaper/easier/faster than current designs. No doubt many customers will see value in these new ways of doing things, and everybody will benefit.
But that’s not what I find fascinating about SDN. What I am starting to see are ideas that are completely out of the box, and would likely not be thought of by typical network technologists working alone. Let me discuss a few categories of things I have seen possible with the emerging technologies.
The Network as a Compute Resource
It turns out that for certain functions purely within the domain of computing, there are huge advantages to be gained by bringing the network inside of the computer architecture. I have been shown, by one innovative start-up, that by using utilities within the network it could be possible to increase performance by 100s if not 1000s of times. Amazing … and totally enabled by SDN.
SDN as a Service
Why re-buy what you already own just to get the benefits of SDN? That’s the theme behind one new company that is creating a service that can deliver many of the benefits we find in SDN, but deliver them as a subscription service out of the cloud to your existing equipment. If it works for you, keep your monthly subscription -- if not, try something else. Brilliant!
New Middleware Providers
In today’s networking the concept of “end-to-end” often falls apart as soon as you cross a WAN access boundary, or your data moves between different service providers who can do no more than interoperate at the peer protocol level. But what if a new class of providers could see above and beyond those boundaries, and they were able to create a business model in which everybody benefited from their service, including the existing network service providers? Now we’d be talking!
The common thread between all of these is that they combine technology innovations and business model innovations. Interestingly, we are seeing business models from other domains becoming applicable in the networking domain because of the way in which the network is opened up by APIs.
Personally I don’t know any more about NfV or control plane / data plane separation than I did 6 months ago … but I am more interested in the power of SDN than ever before. To me, yea, SDN changes some things about the technology … but it can potentially change even more about the business of networking. And that’s really fascinating.