For an inside look at the goals, expectations and plans of the ONF's recently created Migration Group and how it hopes to help companies make the transition to SDN, Network World Editor in Chief John Dix caught up via e-mail with the chair of the group, Justin Joubine Dustzadeh, CTO & VP of Technology Strategy, Networks, Huawei.
When was the ONF Migration Group formed and what is the goal?
The ONF Migration Working Group was formed in April of 2013 with the main goal of producing methods for migrating network services from a traditional network, such as a data centre network or wide area network, to an OpenFlow-based software defined network. Our initial focus has been to examine real-world software-defined networking (SDN) migration use cases, gather best practices and make recommendations on migration methods, tools and systems. Our broader objective is to help the network industry accelerate the adoption of open SDN.
How did you get involved?
Open, standards-based SDN is a key component of Huawei's SoftCOM vision for the evolution of carrier and data centre networks. Huawei has been a strong advocate and supporter of OpenFlow since the early days and we have been contributing to various activities within the Open Networking Foundation. As part of my day job as Huawei Networking CTO and VP of Technology Strategy, I often interact with customers to discuss their network evolution vision (including SDN) and migration plans. One frequently asked question is about the feasibility of a full transition from existing networks and systems to an OpenFlowbased SDN environment. This question can be particularly relevant in wide area networks due to the greater complexity and potentially more significant infrastructure investment required. Because of my previous operator background (both in the U.S. and France) and my experience with network transformations, I volunteered to help when the ONF leadership team decided to create the Migration Working Group. It has been a great experience.
Where do you stand putting the group together?
The group has been pretty active and the level of interest seems to be steadily growing. We have active members representing network users and operators in U.S., Europe and Asia, as well as folks from the vendor community. There has been a great amount of valuable contributions from operators in particular. I am optimistic about the future of the group and what we can deliver over the coming months.
What are the milestones you are shooting for and the timing for those?
There are currently four milestones defined in the Migration Working Group's charter. The first one, which we just finished, is a document describing key SDN migration use cases and methods, leveraging the experience of prior work by network operators. This document also provides a framework to apply migration methods to a range of target networks, including network core requirements, starting network migration requirements, phased migration requirements and, finally, validation requirements to ensure a complete and successful network migration. The second milestone focuses on specific metrics necessary for a successful SDN migration. As part of the third milestone, our goal will be to publish working code for validating these migration metrics. And our fourth milestone (currently scheduled for April of 2014) will be a demo of a prototype migration tool chain, including simulators and software maintenance capabilities.
Some industry players say SDN is the top level category and network virtualisation a subset, while others say network virtualisation is not SDN. How does your working group view that? What distinctions do you draw?
To date, our group has not discussed this viewpoint specifically. The way that network virtualisation is implemented makes a difference, and could make it only a subset of SDN. My personal opinion, though, is that OpenFlow-based SDN offers more than just network virtualisation.
Will the group identify specific SDN use cases and/or different migration strategies?
Yes and yes. One of the important observations is that enabling (new) services is often the key motivation for SDN migration. These services can be end-to-end, overlay on virtual networks, spanning several network segments and/or across several layers of networking technologies, some or all of which could possibly be addressed by OpenFlow -- and as OpenFlow is evolving, it could possibly address Layer 0, 1, 2, 2.5, 3, 4-7 applications. As such, we believe there is great diversity, and there is value in examining use cases from different network types to better understand unique migration strategies, tools and methods which could be specific to each service or network type. To date, we have examined the following three use cases: Campus Network use case: Stanford OpenFlow deployment Network Edge use case: NTT's BGP-Free Edge field trial Inter-Data centre WAN use case: Google's SDN-powered WAN (B4)
Personally, how do you think companies will adopt SDN ? Will it involve finding niche opportunities and growing from there?
Market acceptance and adoption of SDN will continue to be driven by specific use cases that solve real problems for the network users and network operators. Like with other disruptive technologies, SDN adopters must implement SDN solutions that bring tangible business value -- for example, reducing complexity and cost, and/or creating new revenue opportunities.
SDN vendors should also be engaged in the process of developing a well-thoughtout plan for SDN migration, with the right methodologies, tools and systems in place to help network users and operators make the transition worthwhile from both a business point of view as well as from a technical feasibility perspective (e.g., ability to migrate existing services with equal/better performance without impacting the end user's experience).
With SDN adoption accelerating in the data centre/enterprise space, OpenFlow-based SDN is no longer considered a future "bleeding edge" technology for carrier networks. To date, a number of successful live trials have been implemented (Huawei has done a few customer trials), offering real opportunities for carrier SDN adoption. How long before SDN begins to drive significant industry change?
SDN has already made a significant, irreversible and positive impact on the world of networking, on both the data centre/enterprise and carrier sides. It is refreshing to see the network industry going through a rejuvenating, long over-due transition toward software as it works to catch up with the IT industry. This is a time of great opportunity that we cannot afford to miss.
To maximise the benefits of such a transition and to fulfill the true promise of software-defined networking, we as an industry must build and adopt open,standards based SDN solutions, and not allow a possible regression toward the old paradigm of proprietary interfaces on top of closed network equipment environments (read: vendor lock-in). Moore's law, continued technological advances with distributed systems, cloud/virtualisation orchestration and automation, and a thriving startup ecosystem will make the adoption of open SDN a reality muchsooner than some might expect.
Get our full SDN package: