The Evolution of Open Networking to Automated, Intelligent Networks


The 2018 Open Networking Summit is happening this week in Los Angeles. Just prior to opening day, we talked with John Zannos, Chief Revenue Officer at Inocybe, to get his view on the state of open networking and changes in the foreseeable future. Zannos, is on the governing board of the Linux Foundation Networking effort, and has formerly served on the OpenStack and on the OPEN-O boards.

Inocybe has been involved with OpenDaylight since the beginning. The company is one of the top five contributors, and its engineering team is involved in helping solve some of the toughest questions associated with SDN and OpenDaylight. For example, company engineers lead the community effort focused on solving the problems associated with clustering, security, and service function chaining.

John Zannos, Chief Revenue Officer at Inocybe

Previously, Zannos ran Canonical’s cloud platform business and helped drive the NFV and SDN strategy within the company.  “I have seen the evolution of disaggregation, automation of open source in compute and we are seeing those same elements migrate to the network,” he said. “And, that’s what I thought we should talk about — how SDN and open networking are combining to deliver the promise of automated and intelligent networks.” Here are some insights Zannos shared with us. What is the state of open networking now?

John Zannos: Open networking is here now. Over the last 10 years, there has been open source in the compute space: Linux, virtual machines, OpenStack, Kubernetes. We learned a lot over those 10 years and we are bringing the experience and hard learned lessons to open source in the network.

In the networking space, we have seen NFV as a way to bring virtualization to networking. And we are at a point now that there is leadership from large service providers like AT&T, China Mobile and Deutsche Telekom, and smaller ones like Cablevision in Argentina to name a few. Different members of the vendor community, like Nokia and smaller ones like Inocybe, are navigating how to incorporate open source into the network in a way that it helps accelerate end user adoption with service providers and enterprises, with the goal of achieving the end state of an intelligent and automated network.

At Inocybe, we are accomplishing this through our Open Networking Platform. The Open Networking Platform simplifies the consumption and management of open networking software such as OpenDaylight and OpenSwitch. It helps companies consume just the right amount of open source components they need for specific business use cases (ie Traffic Engineering). We create a purpose built open source software stack that is production-ready for the specific use case. It helps organizations automate the build, management and upgrade process, ultimately putting them on a path to automated and intelligent network.

At Open Networking Summit, we’ll be demonstrating how our Open Networking Platform can deploy a fully integrated OpenSwitch-based NOS and OpenDaylight-based SDN Controller on a variety of hardware platforms, eliminating the complexity from the controller down the stack, while preserving the ability to disaggregate the solution (Dell’s booth, number 43). What are the evolutionary steps taken, and still ahead for Open Networking?

Zannos: The first step of this journey was disaggregation of network appliances, separating network hardware and software. The next step was to incorporate automation. An example of that is the use of SDN controllers, such as OpenDaylight, an open source project which automates the deployment and management of network devices.

The next two steps are a combination of data analytics and machines learning/AI. We are moving from collecting data to determine what is happening in the network and what will happen next, to machine learning/AI that will consume that information to determine what action to take. With these two steps we move from analysis to action to autonomous networking. We see open analytics projects like PNDA, which is part of the Linux Foundation Networking effort, moving us in this direction.  In the machine learning and AI space, AT&T and Tech Mahindra with the Linux Foundation have announced Acumos, which will enable developers to easily build, share, and deploy AI applications.

Ultimately, we are using collaborative innovation to help service providers and enterprises be able to use automated and intelligent networks quicker. What’s interesting is that open source creates a framework for companies that compete to collaborate and share information in a way that accelerates adoption to an intelligent, automated network. We are now at a point where we are starting to see those benefits.

Think of software-defined networking (SDN) as allowing for automation and flexibility, and open networking as allowing for collaborative innovation and transparency. When you combine SDN and open source networking you begin to drive the acceleration of adoption.  You said the open networking community could learn from open source adoption in the compute space. What are those lessons to be learned?

Zannos: There are two things to be learned from the compute experience. We don’t want to create too many competing priorities in open networking, and we want to be careful not to stifle innovation. It is a tricky balance to manage.

There was a moment in OpenStack that we had too many competing projects and that ultimately diluted the impact of engineering resources in the community. We want to ensure that the developer and engineering resources that companies big and small bring to the open source communities, can stay focused on advancing the code base, in way that helps drive end user adoption. Competing priorities and projects can create confusion in the marketplace, and that slows down adoption. Companies weren’t sure if all these projects were going to survive. I believe we have learned from that experience. We are trying to be more thoughtful about helping projects form with a focus on accelerating time to adoption by end users where they can actually reap the benefits.  That’s exactly what we are trying to do with OpenDaylight, let it to continue to evolve, but also let it stabilize so customers can actually use it in production.

The second thing is to be sensitive of the fact that you don’t want to stifle competition. You do want to allow for innovation that comes from different and competing ideas. But, I think we have an opportunity to learn and improve from our experience to date.

I am optimistic that our experience as an industry and a community is building a strong foundation for open source adoption in the network. It is exciting to be part of what Inocybe and The Linux Foundation are doing in networking, because it’s an opportunity to collaborate and prioritize the efforts that will help drive adoption.

This article was sponsored by Inocybe and written by

Sign up to get the latest updates on ONS NA 2018!