The Open Networking Foundation (ONF) recently unveiled plans to redefine itself for the post-hype phase of software-defined networking (SDN). I welcome the ONF’s reset and believe it bodes well for the industry’s future.
When OpenFlow Was King
The ONF was born six years ago as a nonprofit consortium charged with the standardization and certification of OpenFlow, the SDN protocol that enables the decoupling of the network control plane and data forwarding plane so that network operators could centralize the control plane in a controller for improved network programmability.
OpenFlow enjoyed a tremendous amount of hype in its early years. Remember when network hardware vendors boasted about how many of their switches supported it? But that hype never translated in significant adoption. In fact, two of the most popular SDN solutions on the market today (Cisco ACI and VMware NSX) make very limited use of OpenFlow, if at all. Despite this reality, the ONF’s core mission continued to be OpenFlow development and standardization, and it maintained a puritanical definition of SDN, insisting that any SDN architecture must have control and data forwarding plane separation.
Early SDN Adopters Don’t Care About OpenFlow
EMA surveyed early adopters of SDN more than a year ago and found that OpenFlow and the decoupling of the data plane and the control plane were not important to network operators. We asked them to identify the defining characteristics of SDN that were most essential to the solutions they implement. Out of 11 multiple choice answers, OpenFlow placed 9th and data/control plane decoupling placed last. Instead, early adopters valued solutions that provided (1st) a centralized controller, (2nd) low-cost hardware, and (3rd) network agility.
In other words, network operators don’t care about which protocols a solution uses or the architectural philosophy of a given vendor. Instead, they focus on the beneficial characteristics of a solution. They want centralized control, but they don’t care how a vendor provides it. Well, that’s not entirely true. Every technologist cares about how a solution works, but they aren’t necessarily going to kick a vendor to the curb just because it enables SDN by some means other than OpenFlow and the architecture promulgated by the ONF.
By staying laser-focused on OpenFlow, the ONF risked irrelevancy. Fortunately, things began to change last October when the ONF announced it was merging with Stanford University’s Open Networking Lab (ON.LAB). At the time, ON.LAB was best known for developing the open source Open Network Operating System (ONOS), a highly-scalable SDN controller in various stages of trials and adoption by leading network service providers. Last year ON.LAB handed control of the ONOS project to the Linux Foundation, which has become the home of roughly a dozen different open source networking projects.
The ONF’s New Mission: Solutions
The ONF announced this month a new mission for itself. While it will continue to be the steward of OpenFlow, its central focus will be on creating and leading an “open innovation pipeline” for open source networking. In other words, the ONF will try to marshall together a community of network operators and vendors and lead them in the development of an integrated stack of open source networking technologies. There are a multitude of open source networking projects out there today, all working at different layers of networking, from data forwarding to network control, all the way up to the service layer and the orchestration layer. The ONF will coordinate the development of specific solutions that pull together these different technologies.
This innovation pipeline will focus initially on solutions in the network service provider world, specifically a handful of different flavors of CORD (Central Office Re-arhitected as a Datacenter). In a few months, the ONF will announce a roadmap for similar plans in the enterprise networking space, so stay tuned to that.
In the meantime, it is clear that the ONF aims to claim a new leadership role in the SDN and open networking world. I welcome this evolution. SDN and open networking are still looking for mainstream adoption. Most organizations are still taking a wait and see approach. The ONF joins a growing community of network operators, vendors and open source champions who are moving the industry forward, slowly but hopefully surely.