05 February 2015

“Constraints that Deconstrain” SDN-NFV

We are witnessing very large but fragmented efforts worldwide in defining SDN and NFV. Very large as almost anybody is realizing that this is not just a new technology (like SDH or IP) but it is "tipping point" for Telcommunications and ICT ecosystems. It's software, it's OPEX-based models. It's much less investments to deploy a new infrastructure (virtualised from L2 to L7 and executed with Pay-as-you-go biz models). It's about many more Players entering the arena. It's changing the value chain, dramatically. So it's highly fragmented by different interests. Waterfall innovation and CAPEX-based models belong to the past.  

All Standardization Bodies, Fora, Projects and Initiatives dealing with SDN and NFV are debating more or less about the same well-known issues and questions, very often with different languages and background (from IT and From Telecom). And everybody seems looking for the winner-take-all SDN-NFV architecture, but in most cases discussions are dealing with conservative perspectives, slow-moving standards, in front of this global "game changer". This is risky, as in this context, "diversity" is accelerating.

My take is that it's unlikely that one of these Body will found winner-take-all solutions, and that standardization will reach fast agreements on certain "strategic" interfaces. One more thing, is that today, there is no formal theory nor systematic design principle for such a network architecture even more when this is more and more appearing as a “borderless continuum” of logical resources (e.g., containers) dynamically allocated on an underneath physical infrastructure.

Rather - quoting Prof. John Doyle - we need focussing on those "constraints that deconstrain" in this case SDN-NFV (i.e., defined by John as that minimum set of constraints that free up design choices everywhere else): but that’s indeed the Operating System, we need to make SDN-NFV to fly.

And to achieve that seriously, we need also following an approach with in line with the times: waterfall innovation model is failing today. That model was moving from research activities to standardization, from systems development by Technology Providers to Network/Service Providers deployments up to the provisioning communication services to Consumers. This is the past.

Winning innovation tomorrow is likely working the other way round: it will start bottom-up, from the real needs and requirements of the Digital Economy and Society, from the behaviour and dynamics of massive Consumers and Producers, from Communities and Social Networks etc. Network architectures are disappearing into the reality as they are become their nervous system of the reality. 

So let's start from concrete use-cases posing real development problems. And real market will decide, adopting also standards-de-facto, if the case

Brittle and fragmented activities adopting a slow waterfall innovation model will not find the way; what’s needed, in my opinion, is a massive joint effort for a new Operating System for Softwarization with a minimum set of constraints required to free up Virtual Functions (i.e., applications) on one side and the physical Hardware on the other one.

So let's start from available test-beds and field-trial aggregating effort towards a coherent global goal. Join our technical community ! Banner on top right of http://sdn.ieee.org/