On April 29, I spoke at the big Interop networking trade show in Las Vegas on one of my favorite topics: the future of SDN. Rather than wearing my Infoblox hat, I took the stage in my role as the chairman of the Open Networking Foundation’s new Software Leadership Council. I suspect ONF gave me the council chairmanship to quiet my dissenting voice within the larger organization, but I was still happy to speak about the massive changes that I believe SDN will bring and to encourage the audience to participate in ONF’s new community devoted to Open Source SDN.
Despite the fact that I don’t always agree with the ONF, I continue to believe that it’s the best hope for the SDN community. I don’t know of any other organization focused on SDN that has such a rich mix of networking consumers and telco operators, not just vendors. These networking consumers include Google, Facebook, Yahoo and Deutsche Telekom, companies that have had to change the way they do networking because the economics of their business demand it.
The ONF wants to create what we need and don’t yet have – a common definition of SDN so we can drive visibility of SDN in the marketplace. If you’re a believer in SDN, ONF deserves your support, especially as hype begins to diminish and real use cases start to appear.
One important thing the ONF has done over the past few months is to commit to being a curator for software artifacts for SDN, and not just at the low technical level such as the controller. Almost all of the discussion in SDN has been at the controller level. That’s like talking about open source operating system kernels all the time. No doubt it’s an important discussion, but there’s not likely to be any differentiation over time. That’s what the new Open Source SND community is for.
If you have new applications – like taking advantage of Hadoop in the data center, or building cybersecurity networks overnight for data processing, or the Internet of Things – this is what SDN brings you. That’s what I believe companies such as Facebook, Goldman Sachs, and others are deploying SDN. They had a driving need for a different kind of economics. That’s where the value is for the marketplace – answering the question of how the economics of SDN will change the way enterprises and operators consume networking technologies and make them function better organizationally. That’s where SDN will really make its impact.
Still think SDN is a fad? Consider this: Standards from the Internet Engineering Task Force made a whole lot of sense when vendors had to develop a custom ASIC processor for their equipment. That could take as long as three years, so there better be a well-defined standard to work from. But if you’re running on a general-purpose CPU and the guts of that networking function are in software, you can upgrade it overnight. I would present that to you as concrete evidence that this paradigm shift is inevitable. It may take up to ten years, because of friction in the marketplace, but I’m convinced it will inevitably happen.
Here’s how I’m sure SDN is fundamental and not a fraud. It’s not just from talking to many Infoblox customers about their needs. I did an experiment at home. I bought a $300 server 18 months ago, one using an Intel Atom processor – nowhere near the top of the line. I installed a programmable soft switch to track the Ethernet packets in and out of my house – yes, this is what I do for fun. But it showed me where my three kids were surfing on the Internet. They all do streaming video, and not one of them noticed that I was sniffing the network. If I had more kids, I might have to upgrade, but the point is that this little software application was able to conduct deep-level insight into network traffic on a low-level server without any impact.
That’s the crux of SDN. It’s not about performance; it’s about performing well enough for what you need at the right price. Now apply that to cybersecurity, transactions, you name it. SDN is not just a flash in the pan, but fundamental to how the networking will evolve over time. Ignore it at your peril.