Re: Defining SDNI could not agree with you more. Not sure why HR felt the need to come up with a definition. Nick's definition adequately covers it--and it's pretty much the way most of the industry views SDN. Leave well alone...that's good advice in this instance.
Re: Evolution, EvolutionI do to, Ray, and I think that's one of the values of something like you're doing here. We need to be aware of how all our technology revolutions are defined--how they were first viewed and how that view evolves. Otherwise we may be seeing the realization of a concept that's changed so much from its inception that it can't address the same issues any more!
Re: Evolution, EvolutionThanks Tom - very useful additions. For sure there are different approaches to achieve the same generaln goal, depending on the network and business case.
I think it is useful to regularly reflect on what SDN and NFV are in relation to the industry and it's healthy to see that there are differing viewpoints (see other messages on this thread).
Re: Defining SDNHow can a few months outstrip a definition? A "table" is well-defined even though there are so many more variants of table than the caveman days. A definition that doesn't last more than a few months only means that we don't know what we are are talking about yet.
Tom's examples only serve to emphasize that we know what we want to have SDN cover, but we haven't yet found a good over-arching definition for it.
Nick's definition sounds more like a property than a definition.
Re: Defining SDNI think that's way too simplistic of a definition. We've had separated control/forwarding elements for decades now, certainly wouldn't call it "SDN."
Tom has nailed a real-world definition. I would also include application driven networking as part of the definition as well. Doesn't necessarily need a separate control and forwarding plane.
Evolution, EvolutionI think that for both SDN and NFV, market evolution has already outrun the traditional definitions. In the case of SDN, we see three distinct models of SDN presented in the market, one truly centralizing the contorl plane, one building an overlay on current connectivity, and one adding software control to current behavior. In NFV we see operators focusing more on operations savings and "service velocity" benefits than on capex reduction, which necessitates an examination of how NFV and legacy network elements coexist, cooperate to build new services, and are managed. It may be that the development of these two definitions will be a guidepost in tracking what the market really thinks about both technologies.
I think it is useful to regularly reflect on what SDN and NFV are in relation to the industry and it's healthy to see that there are differing viewpoints (see other messages on this thread).
Ray's (i.e. HR's) "definition" is jabbering about assumed benefits and says nothing about technology. Tom's comment is on "how", not on "what".
T.
Tom's examples only serve to emphasize that we know what we want to have SDN cover, but we haven't yet found a good over-arching definition for it.
Nick's definition sounds more like a property than a definition.
-desi
Tom has nailed a real-world definition. I would also include application driven networking as part of the definition as well. Doesn't necessarily need a separate control and forwarding plane.
SDN is the separation of the control plane from the forwarding plane..
Everything else is a benefit that derives from that separation
- Nick McKeown, OFC 2013 Plenary