An Update on OpenDaylight

As stated on their Web page, the OpenDaylight consortium is a community-led, open, industry-supported framework for accelerating adoption, fostering new innovation, reducing risk and creating a more transparent approach to Software-Defined Networking.  I recently had a chance to talk with Neela Jacques, who is the Executive Director of OpenDaylight.  Neela started the interview by saying that ninety to ninety-five percent of end users are crying for SDN but that to date there has been very little implementation of SDN because users are forced to make a bet on one of a variety of proprietary solutions.  His take is that the OpenDaylight consortium will be a boon to SDN deployment in part because the open source code they produce will be the starting point that many companies will use to build their SDN solutions.

I will admit that I am bullish on SDN and I also believe that over time that OpenDaylight will be a major driver of SDN deployment.  I also totally agree that there has been very little deployment of SDN to date.  I do differ slightly from Neela on two things.  One is that while I certainly see a lot of end user interest in SDN, I don't see that ninety to ninety-five percent of end users are crying for it.  Also, while there is no doubt that the breadth of proprietary solutions is an impediment to SDN adoption, right now it is only one of several impediments.

Neela admitted that when OpenDaylight was formed back in April of 2013 that there was a lot of skepticism about whether or not it would be successful.  He estimated that back then that perhaps 25% of IT professionals thought that the consortium would be able to "change the world."   However, in February 2014 OpenDaylight issued their first released of code - called Hydrogen.  Neela stated that releasing code has given them creditability and that now roughly 50% of IT professionals believe that the consortium will be able to change the world.  Some very recent events add to their creditability.  This includes HP announcing that they are upgrading their membership in the consortium to the platinum level and Extreme Networks announcing that they will release an SDN controller based on the OpenDaylight software.

The OpenDaylight consortium currently has 9 platinum member, 2 gold members and 28 silver members.  Unlike the typical standards committee, the members of the consortium have a lot of skin in the game.  For example, platinum members commit to dues of $500,000 a year for two years and to also provide at least ten developers a year.  The financial commitment for Gold and Silver members is determined by a sliding scale based on the company's revenues.  Gold members pay annual dues that range between $50,000 and $250,000 and provide at least three developers while Silver members pay annual dues that range between $5,000 and $20,000 and provide at least one developer.

An increasing amount of the collateral that the OpenDaylight consortium creates mentions Network Function Virtualization (NFV).  Neela said that both currently and on a going-forward basis, the consortium will focus primarily on SDN and not as much on NFV.  He added, however, his belief that NFV will be a key driver of SDN adoption because without SDN, NFV is "just not that interesting" and that the NFV use cases that service providers really care about require SDN.

One of the things that I really like about the OpenDaylight consortium is the pragmatic view they are taking.  Nella's position is that the traditional approach taken by standards committees is appropriate for functionality that is relatively narrow in scope and that is relatively well defined.  For functionality such as SDN that doesn't fit that description, he believes that you need to implement the functionality first and then either develop standards around what was developed or use defacto standards.  As part of their pragmatic approach, the OpenDaylight consortium is working on a number of POCs (Proofs of Concept), including ones focused on resiliency and ones focused on traffic engineering.

While justifiably proud of what the consortium accomplished when they released Hydrogen, Neela pointed out that "Only an idiot would take Version 1.0 of an open source code base and use that without modification in a production network."  With that in mind, I asked him what we could expect out of the consortium over the next twelve to twenty-four months.  One thing he suggested is that we will see a growing number of announcements about companies building products based on the software the consortium produces and that this will cause a snowball effect in terms of interest in and usage of the software.  He said that the consortium knows that they need to both make the software bullet proof as well as improve its performance and scalability.  In terms of specific functionality that we may see either this fall in the next release of software (Helium) or sometime not long after that, he mentioned service chaining and the federation of SDN controllers that are based on their software.  He also mentioned the possibility of additional network virtualization options as well as more L4 - L7 functionality.