15 May 2015

A Universal Platform for Softwarization

There are a lot of reference architectures looking for the Softwarization of Telecommunications infrastructures. In my opinion too many, given that in the medium-long term will make no sense distinguishing SDN, NFV...and related declinations. 

Just to mention some of these references: the SDN architecture from ONF, the NFV architectures from ETSI, the architecture from OpenDayLight, the one from OpenStack extended to become an Orchestrator...plus all the reference architectures coming from other RT&D and Innovation projects and initiatives worldwide (e.g., H2020 in Europe). And the number is growing daily, through several other emerging proposals from start-ups…

Will market decide with a standard-de-facto, by promoting a “winner takes it all”? Not sure, in the medium term, there is too much fragmentation, lack of interoperability… 

It’s also true that tomorrow, rather than on thousands of closed equipment, the Telecommunications infrastructure will be based on millions of software processes…executed in distributed logical resources. Nevertheless this does not justify the current fragmentation…this chaotic rush.
This fragmentation is delaying - if not jeopardizing - an effective exploitation of the enormous innovative potential of Software Defined Infrastructures, and their socio-economic impact. Technology maturity and cost reductions are enabling this transformation.

Let's exploit it by creating the foundations of new open ecosystems. Specifically, we are missing coherent efforts in defining a sort of "universal functional model" for SDI, and, in turn, the sw and system architectures which will have to become the standardized references for implementations and exploitation. Should a coherent effort be jointly applied in this direction, there would be an “explosion” of new developments and growth in several ecosystems, with far reaching positive impacts from the socio-economic viewpoint.

This is my take today: let's join a coherent industrial-academic efforts in developing new analytic ways of modelling, designing, deploying, and operating SDIs, in a word, not ten but one universal functional model (a sort of new G.805-like model for SDI) with well-defined functions, modules and interfaces. And it should be able to unify the abstractions of both network and IT services, and the way they are modelled and represented. Based on that, then, we’ll have to design the system and software architectures and all the required processes, which will have to be highly dynamic and flexible…to cope with future ICT markets.