How will Open Source Impact the Development of SDN and NFV?
Another blog in the series by Dr. Jim Metzler
Until recently, open source has had a relatively modest impact on networking. One of the many indications that open source is now having more of an impact on networking occurred at the 2016 Open Networking Summit in Santa Clara, CA. In past years, the focus on the ONS was SDN and NFV. While those were important topics at this year’s conference, the tag line for the conference was “A New Era of Open Source Networking”. The fact that this year’s ONS had such a strong focus on open source shouldn’t be surprising because in November 2015 the ONS became a Linux Foundation event.
A more important indicator of the growing impact of open source is the breadth of open source solutions. For example, one speaker at the ONS conference presented a model of open source software solutions that reminded me of the OSI model. However, whereas the OSI model has seven layers, this open source model has 10. I listened to this speaker both as an optimist and a pessimist. As an optimist, I was impressed by the breadth and depth of open source solutions. As a pessimist, I was struck by how complex it will be to integrate all or some of the 10 layers into a complete solution.
As alluded to by the ONS speaker, there are multiple open source activities that focus on creating an SDN controller. One of these activities is driven by the ON.Lab, which is a non-profit organization founded by people from Stanford University and UC Berkeley. One of the ON.Lab’s primary projects is ONOS (Open Network Operating System). Two of the goals of ONOS are to build an open source SDN operating system for service providers and to build open source SDN and NFV solutions.
In October 2015 the ONOS community and the Linux Foundation announced a partnership that the announcement said will advance and bring together the community for ONOS to build open source platforms, solutions and ecosystem for service providers to monetize SDN/NFV, while helping vendors and service providers invent new business models. In March 2016 the ONOS community announced the availability of its latest software release, Falcon, which is a part of the March 2016 OPNFV distribution called Brahmaputra.
Another open source activity that is focused on creating an SDN controller is the OpenDaylight (ODL) Project, which was founded in April 2013 and is hosted by The Linux Foundation. The stated goal of the project is to facilitate a community-led, industry-supported open source framework, including code and architecture, to accelerate and advance a common, robust SDN platform and to create a solid foundation for NFV. Towards that end, the ODL project claims that a number of vendors use ODL code as the basis of their SDN products and that its code is also used by the OPNFV platform. In February 2016 ODL announced its latest software release called Beryllium (Be). According to that announcement, “Those who have already deployed ODL will see significant improvements in performance, scalability and functionality with ODL Be. New network services offer clustering and high availability, improved data handling, messaging for transport, greater abstraction of network models, broad management of network elements, and a new GUI.” As of May 2016 the consortium had 50 members: 7 platinum members, 2 gold member and 41 silver members.
In a conversation that I had had with Dan Pitt, executive director of the Open Networking Foundation (ONF), Pitt said that he thought that the growing interest in SDN-related open source projects would accelerate the adoption of SDN. His belief in the importance of open source based solutions is the reason why in 2015 the ONF launched an open source community and code repository called OpenSourceSDN.org. The role of this community is to sponsor and develop open SDN solutions in order to provide greater adoption of open SDN. According to Pitt, the work of OpenSourceSDN.org is complementary and interoperable with work being done by open source organizations such as ODL, ON.Lab and OPNFV.
The open source activity that is most closely associated with NFV is OPNFV. OPNFV is intended to be a carrier-grade, integrated, open source platform the stated goal of which is to accelerate the introduction of new NFV products and services with a focus on NFV Infrastructure (NFVI) and Virtualized Infrastructure Management (VIM). In a conversation I had with Heather Kirksey, OPNFV’s Director, she emphasized that OPNFV’s charter is broader than just building a platform that realizes the European Telecommunications Standards Institute’s (ETSI’s) architectural framework. She emphasized that OPNFV is committed to influencing upstream communities such as ODL, OpenStack, and Open vSwitch and that it is committed to enabling both functional and performance testing.
As mentioned, I am impressed by the breadth and depth of open source solutions. However, open source is not a panacea.
- For example, virtually all vendors will add functionality to the various open source solutions prior to bringing them to the market. As a result, these solutions will not be as interoperable as the trade press sometimes makes them out to be. In addition, if it serves their interests, vendors can slow down the process of developing an open source solution or they can steer it in a direction that favors them.
- I strongly believe that open source offers a lot of potential in those situations in which vendors can identify functionality that is difficult to implement and which does not differentiate them in the market. In these situations it is in everyone’s best interest to shorten the time to market by working together to develop open source solutions. I think that it is very positive that so many of the open source communities seemingly want to work together and I am also pleased with intention of organizations such as the ONF to integrate existing open source solutions.
One key outstanding question is the relationship between the open source communities and the traditional Standards Developing Organizations (SDOs) such as the IETF. One option is that after a group such as the OPNFV has made progress creating an open source reference platform for NFV, that one or more SDOs will establish working groups to create standards for some of the key tasks that are part of the reference platform. However, since SDO working groups have historically taken years to create new standards, another option is that some or all of the functionality that is part of the reference platform will become de-facto standards.