Specifically, I’d been curious about whether Juniper’s new OpenContrail effort will end up competing with the OpenDaylight Project, another open-source effort that involves more vendors and is grander in scope.
According to Juniper, the difference between the two is that OpenContrail is addressing a specific problem, while OpenDaylight is more open-ended, trying to address multiple SDN angles at once.
Contrail, as a startup, didn’t plan on taking its code open-source. Bob Muglia, Juniper’s executive vice president of software solutions, suggested it to the startup a year ago. Ankur Singla, who was Contrail’s CEO and is now with Juniper, says his team didn’t completely buy the idea.
But after Juniper acquired Contrail in November, it became clear that the goal had to be widespread adoption of the code. Offering it as open-source code could be a start. And as Juniper executive Brad Brooks explained to SDNCentral earlier, the company doesn’t believe the money in SDN is going to be in elements such as the controller.
But as you might expect, Juniper doesn’t consider OpenContrail to be competitive with OpenDaylight. Contrail is aimed at network virtualization and network functions virtualization (NFV) initially. OpenDaylight is, in a sense, taking on everything, “and we’ll see over time what set of problems it makes sense for OpenDaylight to solve,” Muglia said at Monday’s launch event.
Juniper is an OpenDaylight member, and it seems likely that parts of OpenContrail, if not the whole thing, could be submitted to the consortium later on, he added. (Note that OpenContrail is exactly the same production code Juniper offers commercially as Juniper Networks Contrail. The difference is that the latter comes with service and support.)
OpenDaylight, for its part, doesn’t see OpenContrail as a competitor. “It’s not uncommon for vendors to open source technologies so that their customers can get ‘under the hood,'” writes Jim Zemlin, executive director of the Linux Foundation, in a note to SDNCentral. “The key, though, to long-term success for an open source project is community support, so we fully expect Juniper to submit the OpenContrail source code to OpenDaylight, after which the OpenDaylight community will review the submission and determine what makes sense.”
Open-Source and Community
Juniper isn’t the only SDN player for which adoption-of-code is of paramount concern. I’m assuming more pieces of the SDN puzzle will go open-source, either through consortia or individual vendors. This doesn’t bother Juniper; Muglia noted that open-source movements tend to start with multiple competing ideas or projects.
Ideally, any open-source SDN projects would share a common backdrop — probably OpenStack — in such a way that any of the options would work within that architecture, Muglia said.
It all sounds great, but there’s a difference between a vibrant open-source community and a piece of free code waiting to be picked up. What I’ll be watching is how well the freely available OpenContrail can build up its community, because their input and enthusiasm would go a long way toward driving adoption.
Other bit open-source projects have worked in the opposite direction; they start with something that’s popular (Unix, for instance) and open that code, or similar code, to the masses, who then augment it in all directions.
Juniper is open-sourcing something that wasn’t a product before, one that isn’t starting with a deep community. OpenDaylight can draw from a base of vendors that have committed to contributing code. Juniper, going it alone for now, might have a chicken/egg struggle if it can’t get its own community ignited.
(Featured image: Muglia and Singla present OpenContrail to analysts and press at Monday’s launch event.)