What is a modern data architecture?

At the recent DTW Ignite in Copenhagen, there was an understandable excitement around AI and the potential to transform telco operations with AI. Alongside that however was a focus on the underlying data platform, and what would be required to deliver AI-based transformation. The data platform has been the subject of several collaboration groups at TM Forum for some time, and most recently the Modern Data Architecture Collaboration Group who have just published a whitepaper on the subject.

A modern data architecture, to quote Tomas Sadauskas of Swisscom, “is an architecture that operates at scale, that is flexible and agile, open and extensible. It’s not just a data warehouse any more, but a series of distributed functions from streaming and data pipeline management, to visualisation, collaboration and machine learning, storage and hybrid cloud operations, and a comprehensive enterprise governance.”

Cathal Kennedy from Telenor emphasised the importance of the data consumer as a point of focus. “Historically, we have concentrated on the data producer, figuring out how to capture and store data from massive sources like the network, and then letting others figure out what to do with it. Pivoting to the data consumer, we are now thinking about AI and autonomous operations in particular, and focusing on the underlying data architecture that best serves those needs.”

Historically, BSS/OSS systems used to serve data platforms, that in turn could be used for network or customer optimization functions, amongst other applications. Moving forward, as AI increasingly drives applications execution, and as the data platform is considered more than just storage, extending into data flow management and ingestion / streaming management, a modern data architecture can be reasonably expected to serve the BSS/OSS architecture, rather than the other way around.

Another topic highlighted by Kim Krogh Andersen of Telstra at DTW Ignite was the end of vertical applications, and the move of the IT architecture to a more open, developer friendly environment. That too is reflected in the approach to a modern data architecture being developed by the group.

Having completed this draft of the paper, the Modern Data Architecture Project is currently focusing on a technical report, a more detailed document getting into the architecture, and a summary of operator views on modern data architectures, comparing approaches. If you would like to participate in the work, and contribute your point of view, please join the project.