There’s been some confusion — in my mind, anyway — about the software-defined networking (SDN) mandates being pursued by the Open Networking Foundation (ONF) and the proposed Software-Driven Networking Protocol (SDNP) workgroup of the Internet Engineering Task Force (IETF). I mean, both invoke the same three-letter SDN acronym, though one is “defined” and the other is “driven.”
You might wonder, as did I, how the two differ. You and I would not be alone, because members of the ONF, the wider OpenFlow community, and IETF participants have grappled with the same question.
Fortunately, the emergent IETF workgroup, which ambled into view only recently, is endeavoring to bring some clarity to the picture. Before we explore its continuing attempt at elucidation, let’s first review what the Open Networking Foundation is all about.
Forwarding and Management Abstractions
According to the ONF, it is a non-profit trade organization whose mission is to promote the development and use of SDN technologies. As the ONF website explains, these SDN technologies embody two basic principles:
1, Software-Defined Forwarding: Forwarding functionality should be controllable by software through an open interface. This can be achieved with hardware that accepts from software a set of <header template, forwarding action> entries, where the designated forwarding actions (such as forward out a particular port, or drop) are applied to packets with headers matching the template (which can contain wildcards). OpenFlow is an example of such an interface.
2. Global Management Abstractions: Networks should support a basic set of global management abstractions upon which more advanced management tools can be built. These global management abstractions might include, for example, a global view of the network, triggers on network events (such as topology changes or new flows), and the ability to control network elements by inserting entries into their hardware forwarding tables.
Until now, the ONF’s focus has been nearly exclusively on OpenFlow, a protocol that allows interaction between a software-based control plane, residing on a server, and the data plane of a switch. That focus could change, as the above reference to “global management abstractions” suggests.
Alas, this is where ambiguity intrudes between the tasks the IETF’s would-be SDNP workgroup might assign itself and the mission and objectives of the ONF. There’s more than a little potential for overlap, conflict, and — at least in my case — confusion.
In a discussion thread on the SDNP birds-of-a-feather (BoF) mailing list this past fall, participants sought to draw lines of demarcation between their efforts and those of the ONF. Subsequent to that discussion, Harry Quackenboss, CEO of LAYERZngn, wrote a blog post on the topic.
Quackenboss wrote that, though ONF’s vision is expansive, its standardization efforts have been confined to OpenFlow and to communication between an OpenFlow-enabled switch and an OpenFlow controller. On the other hand, he says, the IETF’s SDN discussions pertain to management frameworks, data models, and coordination of management across networks. Therefore, the IETF discourse might include reference to OpenFlow as well as to other protocols, proprietary and open as well as established and new.
In the aforementioned SDNP discussion thread last fall, David Meyer, distinguished engineer at Cisco Systems, wrote that the nascent workgroup might . . . .
“ . . . . provide a set of ‘network abstractions’ and APIs that provide programmatic automation (etc) of configuration, management, monitoring, data mining, telemetry, … to network services. This is the case for current control planes and will bet he case for future (OF/SDN or otherwise) control planes. So I claim provision of these kinds of APIs and abstractions (i.e., the goal set of SDNP as I understand it) is largely orthogonal to OF/SDN.”
Paris in the Spring
Notwithstanding Meyer’s bid for consensus, the would-be workgroup has yet to cohere on a prescribed mission. Attempts to identify a problem statement and to propose specific use cases continue. The next session of the SDNP BoF is scheduled for IEFT meetings in Paris in March, by which time IEFT participants are expected to have chosen a name and a way forward.
In the meantime, if anybody reading this post is involved in both efforts and can provide further insight as to how the IEFF mandate will be distinguished from ONF’s charter, I would appreciate your comments below.