AI and automation will change networking
The current network model has reached its limits of development, but how will it change? The answers can be found in the software and in the automation. We talked to the coordinator of the GARR infrastructure department Massimo Carboni.
We cannot talk about automation in Telecommunications without considering the important changes that are ongoing. In the wake of the paradigm-shift introduced by the cloud revolution, networking is evolving too. This is because there is growing need for scalability. There are different reasons behind this request, including the continuous need for larger bandwidth connections, an exponential rise of devices on the network, a larger amount of data to manage and, finally, an ever-increasing user mobility and a need for services to be accessed anywhere in a cloud mode. In this context, the software component, automation and artificial intelligence will be strategic for managing processes and ensuring support for decisions. This will expand our ability to get data and information on the network functioning and to plan developments in real-time.
The current network model is based on a multi-level architecture. For each level there is a corresponding element (fibres, amplification and transmission equipment, IP routers, etc.) strictly connected in a hierarchical way. It is almost a static system, which requires manual configurations, and presents rigid points of demarcation. It works and it is resilient, but this network stability is paid in terms of adaptability. Today, however, we observe a different trend: from a vertical organisation in layers we are moving to a "brick" model, where a service is no longer composed by organising the components in a logical sequence, but by accessing the resources in a direct way. It is a disaggregated model and we can represent networks, apparatuses and network functions as different resources that produce different services when they are combined.