Depending on who you talk to, the OpenDaylight Project is either a rare alignment of competing networking vendors...
working to create open source technology that can benefit everyone, or it is an attempt by incumbent networking vendors to delay and disrupt development of software-defined networking.
OpenDaylight is an open source consortium that will create platform-neutral, open source software-defined networking (SDN) technology, according to Jim Zemlin, executive director of the Linux Foundation, which is hosting the project.
The consortium will produce an SDN controller, protocol plug-ins, SDN applications, virtual overlay networks and northbound interfaces. Founding platinum members include Cisco Systems, Big Switch Networks, Brocade, Citrix, Ericsson, IBM, Juniper Networks, Microsoft and Red Hat, but membership is open to anyone interested.
"The founders of this project actively sought to establish a neutral, open community where no single actor could dominate; where the best code could win and drive innovation forward," Zemlin said. "Anyone can develop and contribute code, get elected to the technical steering committee, get voted onto the board and help steer the project in a number of ways."
Engineers and analysts welcome innovation in open source SDN, but they question the intentions of a vendor-driven consortium.
"We would really love to see an open source version of the controller where maybe it is more like a platform and people can add value to it and still get the money back using plug-ins or additional features," said a network architect at a Web-scale company who requested anonymity.
Mark FabbiVice President, Gartner
"We have two requirements … we want a controller that can control both physical and virtual devices. And we want one [virtual] switch to be controlled by multiple controllers."
OpenDaylight could produce open source software to address these functions, much as OpenStack Quantum has helped with the orchestration of these network elements. Yet, he said, "it seems almost too good to be true."
That's because it's difficult to see how a group of competing vendors will be able to come together and agree on choosing the best technology for an open source community.
"I don't have strong expectations for [OpenDaylight]," said Mark Fabbi, vice president and distinguished analyst at Stamford, Conn.-based Gartner Inc. "This is, in my mind, predominantly an effort to change the [focus of the] discussion around SDN from the innovators that have led the movement and driven the technology back into the control of bigger players who have a much more vested interest in maintaining the status quo."
OpenDaylight's governance: The technical steering committee
The technical steering committee will have control over what technology gets incorporated as core elements of OpenDaylight. While it's true that anyone can be elected to the technical steering committee, the platinum members -- the vendors who contribute the most money and full-time engineers to the project -- will appoint most of the initial members.
Right away OpenDaylight's steering committee will have a Cisco engineer on the board, along with representatives from two strategic Cisco partners -- Citrix and IBM. Yet the board also will also have appointees from Big Switch, the leading OpenFlow-based SDN startup, as well as Juniper and Brocade, two direct Cisco competitors.
Still, Fabbi said Cisco will have an extremely strong influence over the direction of OpenDaylight.
"While the announcement had all sorts of big names and little names, predominantly this is a Cisco-led initiative with the cooperation of IBM and the willing participation of Citrix. After that, everyone else just signed on to keep an eye on what was going on," he said.
First OpenDaylight battle: The controller
As the technical steering committee for OpenDaylight takes shape, observers could see conflicts break out over whose vision should dominate the project. Incumbent vendors like Cisco might fight to maintain the value of their vast install bases, while startups like Big Switch and open source ecosystem players like Red Hat could push for innovation that is heedless of protecting legacy infrastructure beyond the basic investment protection for customers. This conflict will lead to battles over technology and the first battle will be over what SDN controller will be a core technology of OpenDaylight.
Cisco has just open sourced its Cisco ONE controller and is submitting it to OpenDaylight for consideration.
"We've open sourced the Cisco ONE controller -- key components of it -- under an Eclipse public license," said Lauren Cooney, senior director of software strategy and planning at San Jose, Calif.-based Cisco. "We did that in advance of the consortium. We're working with a roster of partners to see if there is going to be a fit for the ONE controller inside of OpenDaylight."
Meanwhile, other open source controllers are already on the market and in wide use among researchers and network operators. Floodlight, an open source version of Big Switch's commercial controller, is a notable example.
"We have the world's largest open source SDN community today," said Jason Matlof, vice president of marketing at Mountain View, Calif.-based Big Switch. "We contributed our controller to the Linux Foundation to be incubated in parallel with Cisco's. We know ours is leaps and bounds ahead of theirs by a factor of two years. We've seen their code. The challenge to the [OpenDaylight] community is to ensure that this is a merit-based organization."
Yet merit is subjective, said Brad Casemore, senior analyst with Framingham, Mass.-based IDC.
"I think vendors will each have their own view of what that means. If you ask a startup what merit means and you ask an established player what that means, you'll get two very different interpretations."
OpenDaylight's choice of a core controller could indicate just what kind of open source organization it will prove to be.
"Why would this group -- that supposedly has the best interest of moving SDN forward -- choose a controller based on Cisco's completely unknown technology that isn't developed in the marketplace over something like Big Switch that has a large amount of commercial deployments and is already a controller that is in an open source environment?" Fabbi said. "There is no logic that would lead you to that choice. That's the first great example where you will see who is driving the technology decisions of this project and what their intention is."
Obviously Cisco doesn't agree with that assessment. Cooney lauded the early success of Cisco's ONE beta program, saying the controller has been running in many customer environments, including some large financial services customers. "We have put together an architecture that we think is solid and is going to meet the needs of customers and developers," she said.
"OpenDaylight is not about building something monolithic, but about building something that is modular and extensible for people to add stuff to or pull pieces out and customize it," added Omar Sultan, senior manager for emerging technology in Cisco's data center solutions group. "The ONE controller was really designed to be highly modular, which is perfect for what OpenDaylight is trying to accomplish; to create a framework and then turn it over to the development community to be able to extend it."
Big Switch's Matlof said the platinum members are "all vendors pursuing their parochial interests. If a vendor gets a block of vendors aligned with it, that can trump meritocracy. We are cautiously optimistic, with an emphasis on cautious.”
Let us know what you think about the story; email: Shamus McGillicuddy, news director