With its first SDN release, OpenDaylight is just out of the starting gate
- 05 February, 2014 23:16
The OpenDaylight SDN project released its first code and drew a sellout crowd to a conference this week, but it will take more than that before the effort can be declared a success.
Backers of the open-source effort announced the release of their software-defined networking platform, called Hydrogen, at the OpenDaylight Summit in Santa Clara, California, on Tuesday. The conference was packed with representatives of networking and IT vendors and some of those who rely on networks, such as service providers.
Participants and analysts applauded the release of the software, which was produced in seven months with contributions from 154 developers, but cautioned that OpenDaylight's impact has yet to be seen. Anyone can download Hydrogen for free, but whether it drives SDN forward will depend on who does and where it ends up.
OpenDaylight launched in April 2013 with Cisco Systems, IBM, Microsoft, Ericsson and VMware among its founding members. The vendors called in the Linux Foundation to host the project, which is designed to combine code from many sources to form an open-source basis for SDN and the related concept of NFV (network functions virtualization). Both of these technologies are designed to do for networks what server virtualization did for computing, making it easier to run networks and add new capabilities to them.
With vendors dominating the membership rolls and Cisco and IBM making big contributions, some critics questioned OpenDaylight's ability to push the new concepts, which threaten to disrupt traditional networking. Rivalries among the members also raise questions about their ability to work together.
Competing network vendors have banded together in the past in groups that had little effect, such as the Network Interoperability Alliance that tried to take on Cisco in the 1990s. But Linux Foundation Executive Director Jim Zemlin held up the Hydrogen release, with more than 1 million lines of code, as proof the naysayers were wrong about OpenDaylight. He compared doubts about the SDN project to early claims that Linux would fail. "OpenDaylight is on the right side of history," Zemlin said.
IBM, not surprisingly, is already using OpenDaylight's software. The company announced an OpenDaylight-based SDN controller, the Software Defined Network for Virtual Appliances, at the conference on Tuesday.
Ericsson plans to use OpenDaylight code as part of its larger SDN offering, said Don McCullough, director of strategic communications for the Swedish company's technology group, in an interview at the conference. He didn't specify when that would happen. Ericsson did take one firm step to promote OpenDaylight on Tuesday, announcing a lab at its San Jose facility for testing out implementations of the OpenDaylight system. The lab will provide a hardware platform on which startups and academic researchers can try out software they develop for OpenDaylight, McCullough said.
On Tuesday, Inocybe Technologies, a small SDN vendor based in Quebec, Canada, also introduced an SDN controller based on Hydrogen.
The quick development of Hydrogen is impressive, said Roy Chua, a partner at consultancy Wiretap Ventures. "This was actually quite well done," Chua said.
But the project still has a long way to go, Chua and others said.
"That's just the beginning," said Nick Lippis, a longtime networking analyst and founder of the Open Networking User Group, in an interview at the event. It's great that OpenDaylight was able to produce its first release so quickly, but it won't be a true success until its code is widely deployed in enterprise and service-provider networks and many developers have built products on top of it, Lippis said. As with Linux, that kind of broad ecosystem could generate a lot of value and create a whole new economy in networking. But it might take seven years to build it, he said.
As powerful networking vendors, Cisco and IBM could both help OpenDaylight and hold it back, Lippis said. By including support for OpenDaylight in its products, Cisco, especially, would be a powerful engine to proliferate the project's technology. But suspicions about the big vendors' roles might keep some would-be participants away, he said. Only time will tell how that evolves, though for now, there's nothing amiss, Lippis said: Access to the project is as fair and open as promised.
The next big challenge will be to get more users involved, said Srini Seetharaman, a senior research scientist for SDN at Deutsche Telekom. Most of the people actually involved in OpenDaylight's activities work for vendors, he said, though he added that the same is true of the Open Networking Foundation, which had Google, Facebook and Yahoo among its founders. Deutsche Telekom isn't a member of OpenDaylight, but Seetharaman participates in the project as part of DT's forward-looking Innovation Center and out of a personal interest in SDN.
Even OpenDaylight's big backers say there is heavy lifting to come. For one thing, carriers demand standards for interoperability, so service providers will need common interfaces so SDN products from different sources will work together, Ericsson's McCullough said. "There's got to be a standardization process," he said.
Parts of the OpenDaylight code itself are missing or fall short, according to Vijoy Pandey, IBM's CTO of network OS. The biggest problem to start with will be synchronizing OpenDaylight-based products with the project's latest code base, he said. In fact, IBM is still catching up to Hydrogen with the SDN product it announced on Tuesday, Pandey said.
In addition, the user experience in terms of deploying and upgrading OpenDaylight's SDN doesn't meet users' expectations yet, he said. "There's a lot of work that needs to be done to make that product-worthy," Pandey said.
Making the software fit vertical industries has to be another area of focus. For example, OpenDaylight needs greater security before it can be used in U.S. federal agencies, Pandey said. "That's something that we hope OpenDaylight will do in the future, but right now it doesn't exist," he said.
The project can't be dismissed, but it's not time for a victory lap, either, one analyst said.
"It clearly wasn't just a press release, and to their credit, the community has shown the ability to get beyond the organizational issues and mixed and possibly contradictory motivations, and do real technical work," 451 Research analyst Peter Christy said in an email interview. "But this is certainly a case where the proof of the pudding is in the eating, and the unanswered questions are who does what with the output and what the commercial impact is. That will take some more time to discern."