Giacomo Bernardi, an Informatics alumnus (PhD Wireless Networking, 2007–2011), is part of the team behind a major breakthrough in data centre network design at Amazon Web Services (AWS), a development now being rolled out at global scale. The new architecture, known as RNG (Resilient Network Graphs), rethinks how data moves through the vast infrastructure underpinning cloud computing. Drawing on ideas from random graph theory, the approach replaces traditional hierarchical network structures with a flatter, more flexible design that allows data to travel along many possible paths. Faster, more efficient, and more resilient networks This shift delivers significant gains in performance and efficiency. According to AWS, the system enables faster data transfer, reduces energy consumption, and lowers infrastructure costs, while also improving resilience by removing bottlenecks and single points of failure.The design is already being implemented across AWS data centres and is becoming the default architecture for many workloads, marking a major step forward in how large-scale cloud systems are built and operated. How AWS’s radical network redesign is forging a more resilient cloud From Edinburgh research to real-world impact Bernardi completed his PhD at the University of Edinburgh under the supervision of Professor Mahesh Marina, where his research focused on wireless networking, including distributed monitoring, network management, and real-world deployment.His work included building and testing experimental wireless networks in the Scottish Highlands - early experience in tackling complex, large-scale networking challenges. Turning theory into practice At AWS, Bernardi and collaborators brought together ideas from computer networking, mathematics, and engineering to tackle a long-standing challenge: how to apply the theoretical advantages of random network structures in real-world data centres.While the concept has been explored in academia for decades, practical barriers—such as routing complexity and cabling—had prevented its wider adoption. The team’s solution combines new routing protocols with purpose-built hardware that simplifies how connections are physically organised, making large-scale deployment feasible. The impact of interdisciplinary research at scale The work highlights the continuing impact of foundational research when applied at scale. By translating abstract mathematical ideas into production systems, the team has created a new approach to data centre design that AWS is now deploying across its global infrastructure.It also underscores the role of interdisciplinary thinking—bridging theory and engineering—in addressing some of the most complex challenges in modern computing. Related links Read more about the work on the AWS website Link to paper: RNG: Flat Datacenter Networks at Scale This article was published on Thursday 18 June 2026
