End-Users Weigh In at the Open Networking User Group
Boston, February 13, 2013 — A massive blizzard unexpectedly rocked New England within a week of the initial meeting of the Open Networking User Group at Fidelity Investments Auditorium in the heart of Boston. To us, it was indicative of the wave of change that Software-Defined Networking (SDN) portends for Enterprise IT, network operators, carriers, and the technology providers that serve them.
About 150 SDN users assembled in Boston for this unique event, explicitly focused on offering end-users a forum to discuss and debate how SDN can improve their businesses and how they can exert more influence on the future of networking. Ten emerging SDN technology leaders showed demos outside the meeting room (from which they were excluded), including ONF members Big Switch Networks, Cyan, Netronome, and Vello Systems.
ONUG Co-Founder and organizer Nick Lippis (of Lippis Enterprises and the Lippis Report) set the tone for the event by asserting key critical success factors to enable the path towards open networking:
- SDN represents the second wave of open networking, following the advent of TCP/IP decades ago.
- End-users are at a critical juncture and must proactively decide to pursue open networking to avoid another wave of vendor lock-in.
- End-users are compelled to think differently to go beyond improving the status quo, and realize the vision for SDN.
Lippis identified the imperatives as multivendor interoperability, vendor neutrality (meaning no proprietary stacks), programmability via a northbound API, and deep visibility and monitoring. He predicted that new business models would emerge, and defined the elements of this “perfect storm” as the advent of cloud and mobile computing, merchant silicon, and SDN. He left us with a warning that the industry is at an inflection point, where the future will be either open or proprietary. ONUG and ONF are united in striving for an open future for SDN.
A series of speakers and panels then proceeded to share their motivations for SDN, including business objectives, existing network challenges, and target use cases. While the use cases were diverse, there were a number of common needs:
- Provide significant benefits to offset the widespread impact for SDN
- Optimize the utilization and performance of their network; one speaker remarked that if their stretch target of 60 percent utilization could be achieved (vs. 25-40 percent today), their organization would be “dancing in the streets”
- Apply best practices refined for server management to networking
- Automate to accelerate applications deployment, new services, etc.
- Significantly reduce the largest line item in the IT budget: OpEx
Among the use cases that were shared:
- Achieve the agility needed for cloud delivery
- Manage and integrate two businesses: managed hosting and cloud services
- Virtualize the infrastructure to serve a diverse set of user needs, policies, and performance with a common infrastructure
- Enable more efficient monitoring and troubleshooting through a “Network Packet Broker”
- Optimize service delivery (especially for video)
- Dynamic and flexible load balancing
One of the more controversial issues discussed centered around whether or not SDN would result in the commoditization of hardware. While some in the audience were seeking significant CapEx reductions in networking equipment, others question how much savings could be achieved. Discussion points included:
- Networking gross margins are too high, especially at Layer 3, but just the threat of SDN has caused some major vendors to reduce pricing based on the threat from SDN.
- While merchant silicon offers the potential for so-called White Box solutions, the complexities of the control and management planes, not to mention performance optimizations, are perceived as major risks.
- The crux of the matter for many was not hardware commoditization, but how to avoid vendor lock-in through open networking.
- Even if initial increases in CapEx are expected, especially when adding SDN capabilities to an existing legacy infrastructure, the long-term expectation is for dramatic CapEx reduction.
In addition to recounting their experiences as SDN early innovators, many speakers were eager to assert their best practices for initial SDN adoption:
- The average number of network switches per IT Ops headcount was orders of magnitude lower than in the server domain (~90 vs. 8,000-10,000).
- Deployers must leave room for innovation; rigid standards are subject to the “least common denominator” effect.
- Target cost reduction should be at least ¼ of the current costs in order to justify the challenges of deploying SDN.
- Organizations of all types must overcome the huge momentum of existing solutions on the organization, including staff expertise, tools, and existing operational processes.
- But, “it is better to jump in than sit on the sidelines”; “just do it and plan to iterate.”
While it was difficult to claim a consensus, a number of themes circulated throughout the day:
- SDN is not a fad but a paradigm shift with widespread implications.
- OpenFlow® has elicited a tremendous amount of interest and is emerging as a key enabler for SDN.
- SDN migration and coexistence with existing network architectures and technologies are an absolute must.
- The benefits must be sufficiently compelling enough to justify SDN adoption considering the broad implications throughout the organization.
ONUG Co-Founder Lippis observed: “It was loud and clear at ONUG that IT business leaders are taking the necessary steps to gain control over their network and cloud infrastructures by demanding open software-defined networking via their power of the purse so they get open networking the way they want it.”
The ONUG Board did not waste any time to discuss the next steps, meeting immediately following a reception held after the meeting.
And finally – in spite of the storm – the streets were clear, sidewalks passable, and business resuming once again. Major forces, natural and otherwise, can be managed, but not without disruption.
-Marc Cohn and Dan Pitt
Marc Cohn serves as the Chair of the ONF Market Education Committee. He is a Senior Director of Market Development at Ciena. For over 20 years, Marc has driven and promoted successful communications software products for the Data Communications and Telecommunications markets as the industry was transformed by the IP revolution. Dan Pitt is Executive Director of ONF. Both attended the ONUG meeting.