At the OpenDaylight Summit in July, there were a number of telltale signs that the project is really hitting its stride.
The membership of the User Advisory Group – a fairly active group – is also pretty impressive:
The second hopeful sign was perhaps my favorite of the general sessions: a panel on “Models of [open-source] Collaboration,” featuring leaders from most of the open SDN projects shown (Prodip Sen was more or less representing an unlisted but related project, OPNFV). Topics covered included:
- How their respective groups work with each other, their interdependencies and occasional overlaps
- The current state and future of OpenFlow
- How “open” a group really is – several of the groups originated from in-house projects at a particular vendor, which were then open-sourced, and the majority of committers often come out of that vendor initially
- Effective models of open-source governance
- The role of research and education networks as incubators, and how their research can be better shared
It was a robust discussion that didn’t shy away from discussing current challenges and points of debate, and to which the audience readily joined in. To me this really spoke to the coming together of interested parties as an active, engaged community – and a healthy one that doesn’t shy away from dealing directly with hard questions in an open forum.
The third item of interest was a breakout session led by OpenDaylight TSC Chair Colin Dixon and Devin Avery. It was titled “Best Practices and Pitfalls for Building Products out of OpenDaylight,” and it attracted a standing-room-only crowd.
Obviously, a few vendors have already done some productization, but it’s heartening to see a broad range of attendees wanting to understand how to do it right the first time. These would include not only potential controller offerings but also ODL-based applications (there were some being shown in the Expo hall already) and perhaps even service offerings. It’s important to develop a common mindset within the community about productization so that offerings from different vendors can be assured of portability and interoperability. This is, after all, why users are gravitating toward open-source to begin with.
So, the key ingredients are coming together:
- Developer mindshare
- Engaged, influential users
- A defined ecosystem of complementary projects (which share a number of developers and users, who all have a strong interest in ensuring collaboration)
- A young but rapidly growing commercial ecosystem with some strong anchors already in place
The roadmap for the upcoming Beryllium release and beyond shows increasing focus on the maturity and reliability of the OpenDaylight platform, reflecting the needs of users moving into production, along with interesting work on new capabilities in a couple of key areas.
One of the “governance” questions that came up on the collaboration panel was one of project focus vs. an anything-goes approach to guiding developer efforts and investments. There has been a jump in the total number of active projects within each release, which certainly says that anyone who wants to propose something reasonable has the opportunity to do so.
At the same time, Dixon and other leaders, in a separate panel about future directions, continued to emphasize meat-and-potatoes efforts around scalability and security, making it clear that their own focus is firmly on ensuring those concerns get sufficient developer support. There’s also quite a lot of largely unsung engineering work going on behind the scenes in the areas of integration and testing. In an open-source project where all participation is voluntary, ensuring support for less “shiny” but materially important work is a matter of leadership, diplomacy, and persuasion.
Confidently making public statements about such things as key roadmap items speaks well to the seriousness of these leaders about ensuring OpenDaylight’s continuing position as a leading general-purpose controller.