OpenDaylight: the next Penguin?

By Scott Bradner

It seems to be passe to be solid these days. The latest example of this is the just announced OpenDaylight Project. (http://www.opendaylight.org/) A bunch of the biggest names in computing and networking have gotten together to push an open source development effort to support software defined, i.e. virtual, networking under the umbrella of the Linux Foundation. This is an impressive group - 18 generally very well known companies have signed up as members so far, split across 3 levels. I expect that the level is determined by how much the company is willing to pony up but I've not been able to find that detail, even in the on-line bylaws. (http://www.opendaylight.org/project/bylaws).

The topic of software-defined networking is well hyped but hard to nail down. The OpenDaylight Project has a good technical overview (http://www.opendaylight.org/project/technical-overview) but even there it sounds more than a little like mumbo jumbo: " The business logic and algorithms reside in the applications. These applications use the controller to gather network intelligence, run algorithms to perform analytics, and then use the controller to orchestrate the new rules, if any, throughout the network."

The first time I heard about software defined networking, at least this iteration of the concept, was in 2009 when in San Francisco with the IETF. At the time the concept seemed somewhat cool but also seemed like a solution in search of problems to solve. If the PR is to be believed virtual computing (a.k.a "the cloud") is such a problem. I'm not sure yet.

But I do know and trust a number of the people of the OpenDaylight Project board. (http://www.opendaylight.org/project/board-members) There are some rather sharp people there and if any group can develop useful technology this seems like a good group to give it a try. According to the schedule, we should know if they pull it off in the fall.

I'm still not quite sure what "it" is though. And, more importantly, I am far from sure how to manage "it" when "it" is deployed. I have always been leery of managing complex virtual environments. (http://www.sobco.com/nww/1996/bradner-1996-08-19.html) Seems like its rather easy to build an environment that is too complex to actually be managed, especially with the current management tools.

The OpenDaylight Project technical overview says: " Software Defined Networking (SDN) separates the control plane from the data plane within the network, allowing the intelligence and state of the network to be managed centrally while abstracting the complexity of the underlying physical network." Two things worry me about this: the central management since the project does not mention developing any tools to do that management; and abstracting complexity since even non-abstract complexity seems to be nearly impossible to deal with when something minor goes wrong.

I do note that the members of the OpenDaylight Project seem to be on the supplier rather than the user side. I do not see Amazon or Google on the list. To me, the project would be more interesting if there were big operators in the mix.

disclaimer: Like most big companies, Harvard is working in the cloudy space (e.g. http://www.mghpcc.org/) and may find that what the OpenDaylight Project produces turns out to be useful but that is too early to say and the university has expressed no opinion on daylight as an objective.