Amazon Thinks the Future of Data Centers Depends on a Technical Problem It Just Solved
Amazon Web Services claims to have solved a decades-old networking problem by deploying a 'quasi-random' design called RNG in its data centers since late 2023, boosting speeds and cutting energy use while eliminating bottlenecks. The breakthrough, detailed in a paper titled *RNG: Flat Datacenter Networks at Scale*, replaces traditional fat-tree topologies with a more efficient, scalable architecture supported by a new device called ShuffleBox, which automates cable shuffling across 20 million kilometers of fiber optics globally.
Amazon Web Services (AWS) has quietly deployed a new networking design called RNG—short for *resilient network graphs*—across its data centers since late 2023, solving a long-standing technical challenge in cloud infrastructure. The system combines structured and random network elements to eliminate bottlenecks, significantly improving data speeds while reducing energy consumption. AWS engineers, including academics, developed the technology alongside a custom device called the ShuffleBox, which automates cable management for the architecture. Traditional data center networks rely on fat-tree topologies, a layered structure of switches and routers that has dominated since the 1980s. While reliable, this design is rigid and inefficient, requiring vast amounts of cabling—AWS alone uses 20 million kilometers of fiber optics, roughly the distance from Earth to the moon and back 25 times. The new RNG approach flattens the network, removing hierarchical constraints and simplifying infrastructure. AWS claims this is the first time such a random-network design has been successfully scaled in real-world data centers. The breakthrough was detailed in a paper published last month, titled *RNG: Flat Datacenter Networks at Scale*. Unlike other tech advancements tied to AI, this innovation focuses on improving core data center efficiency rather than accelerating AI workloads, which have more predictable traffic patterns. The project builds on decades of research, including a 2012 concept called Jellyfish from University of Illinois Urbana-Champaign, which explored random network graphs. AWS recruited researchers from academia to tackle the problem, culminating in a system described as 'mind-bending' in complexity by networking expert Brighten Godfrey, who was not involved in the work. Matt Rehder, AWS vice president of Network Engineering, stated the company is the only one deploying this at scale, offering a competitive edge in cloud infrastructure.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.