17 October 2015

Big (Actionable) Data for Big Networks

Telecommunication and ICT infrastructures are becoming more and more pervasive, hyper-connected and dynamic. Internet of Things, Machine to Machine, advances Robotics, A.I. applications are posing very challenging requirements.

In fact, heterogeneity of nodes, devices and terminals are thus creating a complexity in management and control with is becoming more and more challenging due to the scalability, dependability and real-time action-reaction requirements.  

Industry, as such, is looking with interest at innovative “service oriented” approaches to management-control and orchestration, where infrastructure resources are virtualised and services are becoming sort of “units of orchestration”. Think about the "X-OS".

It’s the transition from a situation where Network Providers are managing-controlling pieces of equipment towards the ability of managing-controlling virtual resources or containers, and orchestrating the various aspects of services.

When automated, this will simplify dramatically the Operational processes, reducing mistakes and delays. It's well known, in fact, the that time-to-market requirements are very critical: any delay in configuring resources and services directly affects deployment and in turn has an impact on savings and revenues.

Some time ago, IETF has released NETCONF and YANG which are standards addressing configuration management: NETCONF is a configuration management protocol for transactions and dedicated configuration operations; YANG is a data modeling language used to model configuration and state data manipulated by NETCONF. YANG can be mapped to a NETCONF XML representation easily.

In other words, interestingly, YANG can be used for configuring both services and the network functions: at the end of the day it’s data, actionable data. This is not new, obliviously, but what makes it very attractive is the vision of the network and service platforms as a "field" of actionable data.

Future management-control and orchestration will be more and more about processing this continuous "field" of actionable data in real time: the "power of events".