Reimagining Networking: A Lofty Goal?

Since the beginning of the internet there hasn’t been something so profoundly different in networking as what’s happening right now with SDN. In the past decade the transformation toward virtualization has been occurring in the rest of the computing infrastructure, but applying these same principles to networking is proving much more challenging. Networking remains today a hardware-centric fabric that must be manually configured and reconfigured. We all know we need to move to new model of IT where humans program the services, requirements and constraints, but yet the hard realities of our current networking model get in the way. We all know we need a major shift in the way networks are architected, programmed and managed but the challenges are also daunting.

This type of innovation in networking hasn’t happened since the internet was born or the web browser was introduced. The Mosaic browser changed everything - it moved us from a point-to-point model to an information fabric, where the value moved from the files to the network. Servers, Tomcat, the Java ecosystem etc., the entire programming paradigm shifted as a result of those activities. That was the introduction of a new platform that was so very empowering. It allowed information to explode but it would not have been possible without the underlying infrastructure and web servers. The browser gets the claim to fame, but without web servers and the ecosystem there would be no Wikipedia, Amazon, Google or Twitter. Leaps in infrastructure innovation are what make game-changing ideas possible.

What will SDN allow us to do? What innovations will we see SDN enable? It won’t just be faster Netflix. We’re going to see an entirely different explosion of ideas emerge because once again we are improving the underlying infrastructure. Networked self-driving automobiles with only a few inches between them increasing road capacity by a factor of 10 while eliminating traffic fatalities; pervasive healthcare where there are more healthcare outlets than Starbucks, where the 80 percent of healthcare that is routine is delivered via the network, supported by on-site nurses with immediate access to both expert systems and every medical specialist they may need; adaptive learning environments that tailor each child’s lesson to both where they are and how they learn best (shoutout to my wife Genia Jacques whose company Kidaptive is at the forefront of this movement); finally the internet of things where almost every physical object provides data that can be analyzed, correlated and acted upon. There are so many innovations waiting in the wings, waiting for the network to catch up to the gains we’ve made in compute. I am often asked why I left a cozy career at VMware to lead the OpenDaylight community - the best answer I can give is that this community has to drive and facilitate innovations that will change the way all of us live our lives. That opportunity only comes around once in a generation.

OpenDaylight Year Two

The OpenDaylight Project is excited to be part of that change. We are one year old today and are still in the early stages of architecting our SDN and NFV platform. We have a lot to cover in year two, for example the OpenDaylight community is discussing the best way to support a wide range of hardware and protocols. How do we evolve the Service Abstraction Layer? Another key question in our industry is about the degree to which we centralize intelligence and how we do this. Are overlays the answer? Is the key simply to pass policy information to super-intelligent silicon? If overlay, what should an overlay look like and what coordination if any must happen between the overlay and the underlay? How do we manage elephant flows? From a controller perspective, how do we parallelize, federate for scale and geographically?

From a more tactical perspective there is always the work of improving performance and stability, and performing interoperability testing. There is also key work to be done with OpenStack and the ONF on standardizing northbound interfaces. I’ve just touched the tip of the iceberg here. The key is that we are internalizing the industry’s biggest debates inside of the community. Rather than people fighting with slideware and ideas (standards bodies) or by implementing competing, poorly interoperable fully proprietary solutions, OpenDaylight is providing a forum for the development and user community to innovate together, building on each other’s ideas (read: code) in a far more constructive manner.

This is certainly a daunting task. I don’t fault the people reading this that wonder if we can do it. I also initially had my doubts, but have had my eyes opened by this community. It is just amazing to observe the magic of people from different companies all huddled around one whiteboard debating these ideas, the next day furiously coding. These are not easy questions nor is there unanimous agreement but that’s a sign of good debates and ideally good outcomes. Your views on this are also highly important and we invite you to join our weekly calls or mailing lists, or apply to join the exclusive OpenDaylight User Group (email me if you’re interested njacques at opendaylight dot org) to share your thoughts. The good thing for anyone who wants to get involved is that we’re just getting started. There’s a lot more foundation to lay and we want your ideas on how we do that.

We also want to help the community grow in a way that is supportive of our major goals like code quality and building an ecosystem around OpenDaylight. For year two this includes:

• A summer internship program to help the next generation of developers get involved.
• Creation of the OpenDaylight User Group (ODLUG) for end users who want to give their input to the community on what they want from SDN and NFV.
• A Design Summit this fall to keep the community talking, innovating and debating.
• A training program for developers and end users to get started using OpenDaylight and contributing code.

These are just a few things on our list. We’re firing on all cylinders and are excited about the year ahead. It’s rare to have such a tectonic shift in industry happening at many levels, but it’s happening right now. There’s so much opportunity in networking to reshape the ways things are done for the next 20 years. We are part of the movement of people who think differently about the future of the network and want to be part of the change.

Take our one-minute survey and tell us your #SDNWish for 2014.