Lessons learned on the path to open, scalable networks

The aim is to build future networks that embrace openness and adhere to cloud native principles, while creating new opportunities for monetization. But it is not simple to do and will involve an overhaul of operational systems and the harnessing of machine learning and AI to drive automation.

CSPs and their partners are already putting in place mechanisms and processes to build and monetize more open and automated software-based networks, including AIOps, Open RAN, network slicing and networks APIs.

Nonetheless as operators piece together the elements that will help deliver future hyper automated networks, they are entering largely unchartered territory. And as they advance, they are learning important lessons about how to create intelligent, customizable networks.

Some of these lessons were shared by network transformation experts from Telefonica, BT, NTT Data, Red Hat, Telecom Infra Project and Analysys Mason during a closed-door session at Digital Transformation World in June, hosted by global systems integration firm NTT Data.

Laying solid data foundations

Given the scale and scope of change underway, speakers from Telefonica underscored the importance of laying solid data foundations. Not only does data fuel AI- and ML-based automation, it gives business owners insight into what is happening at the network, service delivery, and customer experience levels.

A clear vision of what is going on in the network and how it impacts service and customer experience will be essential as Telefonica moves towards its mid-term goal of achieving TM Forum’s level 4 of network automation, and longer term towards a self-X network, moving the company away from a network operated by humans and helped by machine to what speakers described as a “network operated by machines, and helped by humans.”

Telefonica, however, admitted to encountering growing pains and stressed the importance of having a “methodology and framework that allows you to use technology and data. Use case by use case doesn’t work.”

For this reason, Telefonica recommends creating a full data architecture that can offer an end-to-end vision of what is happening from the network through to the end customer.

Any new framework, pointed out speakers, has to be flexible enough to cope with future hyper automation and the evolution of network-as-a-service (NaaS).

Given the complexity of automating and governing the use of data, one of the first steps is to structure the ingestion and integration of data from numerous sources, they said.

Another important measure is to create a data catalog and enable cross-domain correlation so that different operating companies (opcos) can self-serve their data requirements.

“Each opco needs a structure and a framework for deploying use cases.”

Opening the RAN

A more assured use of data and AI and the move towards automation, disaggregation, and cloud-based infrastructure also provides fertile ground for Open RAN.

Open RAN’s reliance on interoperable software and hardware components is an important element of many major operators’ vision of future disaggregated, virtualized networks based on common, open standards. Indeed, speakers noted there are currently dozens of trials of Open RAN underway by CSPs worldwide.

CSPs’ enthusiasm for Open RAN stems in part from their desire to open up the closed RAN market to new suppliers of hardware and software components.

However, multi-vendor deployments also represent an obstacle to the viability of Open RAN deployments at scale. Speakers acknowledged, for example, that multi-vendor sytstems require substantial systems integration. This will call on skills that many telcos lack and will need to source from a third party, thereby eroding potential cost savings.

Speakers, however, emphasized the need to view Open RAN is a tool for future, enabling service innovation across a cloud native autonomous network that stretches from the core to the edge.

Monetizing future networks

Another essential question addressed during the session was how to monetize future networks, with a focus on network slicing and network APIs.

Discussions revealed it is still early days for future network monetization, with the success of both network slicing and network APIs reliant on interlocking advances in automation, standardization, orchestration, the deployment of composable IT and network architectures, and new monetization strategies.

When it comes to network slicing, speakers pointed to increasing enterprise use. However, they also acknowledged that orchestration is complex, and operations will need to be automated if network slices are to be technically and economically viable at large scale.

Speakers also explored the potential for monetizing network-as-a-service through the GSMA Open Gateway initiative, which is supported by TM FOrum's “Operate APIs” . The initiative involves CSPs collaborating to make network interfaces commercially available to aggregators and developers through standardized APIs. Routes to market for network APIs include working with hyperscalers, which have large developer ecosystems already in place.

Early APIs have focused on enabling developer teams and partners to create new intelligent layers of customer authentication, verification, and security within mobile phone networks. But speakers also pointed to the commercial potential for network-as-a-service (NaaS) APIs that serve edge cloud, quality of service, and network slicing applications.

Nonetheless, there was recognition that more needs to be done before a widespread commercial NaaS ecosystem is in place. Not only is the API ecosystem that bridges network and IT infrastructure still under development, operators that want to benefit from NaaS will need both a composable architecture and a pricing strategy.