TL;DR
Amazon has built and deployed large-scale data centers using RNG, a flat network topology that replaces traditional hierarchical designs. This approach improves efficiency, resilience, and reduces costs. The project is now operational in multiple regions.
Amazon has deployed large-scale data centers using a novel flat network topology called Resilient Network Graphs (RNG), marking a significant shift from traditional hierarchical architectures. The deployment, starting in Ireland and expanding to Germany and Spain, aims to improve efficiency, resilience, and operational costs, and is now operational across multiple regions.
Researchers at Amazon, led by principal scientist Giacomo Bernardi and scholar Ratul Mahajan, developed RNG to replace conventional fat-tree networks. RNG leverages random graph principles, achieving high connectivity and resilience while reducing the number of routers needed by 69%. The first RNG data center was built in Dublin in 2024, followed by additional facilities in Germany and Spain in 2025.
Compared to traditional networks, RNG-based data centers deliver 33% higher throughput, cut network power consumption by 40%, and lower operating costs by 27%. The design employs innovative routing schemes like Spraypoint and passive optical devices called ShuffleBox to manage traffic and cabling complexity, enabling scalable deployment within existing hardware architectures.
Impact of RNG on Data Center Efficiency and Resilience
The adoption of RNG at Amazon signifies a major advancement in data center networking, offering a simplified, flat topology that enhances resilience by eliminating single points of failure. It reduces infrastructure costs and power consumption, which can translate into lower costs for cloud services and improved sustainability. This development could influence industry standards and accelerate the shift toward non-hierarchical, scalable network designs in cloud infrastructure.
10Gtek 25G SFP+ DAC Cable – 25GBASE-CR Passive Direct Attach Copper Twinax Cable for Cisco SFP-H25GB-CU1M, Ubiquiti UC-DAC-SFP28, Supermicro and More, 1-Meter(3.3ft), Pack of 4
25GBASE-CR SFP28 to SFP28 Direct Attach Copper Twinax Cable, Passive, 1-Meter(3.3ft) (Note that this length includes two connectors.)…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Evolution from Hierarchical to Randomized Network Topologies
Traditional data center networks have relied on hierarchical fat-tree architectures, which, while scalable, are complex and prone to single points of failure. Research into expander graphs and random networks has suggested potential for more resilient and efficient designs. Amazon’s recent work builds on decades of mathematical and engineering research, culminating in practical deployment of RNG, a flat, random graph-based topology, at scale.
“Build it, test it, and then improve — that’s how RNG became our default for new data centers. The results speak for themselves.”
— Matt Rehder, VP of Network Engineering
TP-Link TL-SG105, 5 Port Gigabit Unmanaged Ethernet Switch, Network Hub, Ethernet Splitter, Plug & Play, Fanless Metal Design, Shielded Ports, Traffic Optimization
𝗢𝗻𝗲 𝗦𝘄𝗶𝘁𝗰𝗵 𝗠𝗮𝗱𝗲 𝘁𝗼 𝗘𝘅𝗽𝗮𝗻𝗱 𝗡𝗲𝘁𝘄𝗼𝗿𝗸: 5× 10/100/1000Mbps RJ45 Ports supporting Auto Negotiation and Auto MDI/MDIX.
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Unverified Aspects and Future Challenges of RNG
While initial results are promising, long-term operational stability, scalability beyond current deployments, and integration with existing infrastructure remain areas for ongoing evaluation. The theoretical models are validated in small-scale or controlled environments, but large-scale real-world performance is still being monitored.
FLYPROFiber- 30M OS2 Fiber Patch Cable LC to LC, Length Options: 0.2m-200m, 1G/10GB Single Mode Duplex, 9/125um SMF Fiber Optic Cable Cord LSZH 30Meter(98ft)
📡High Rated OS2 9/125μm Fiber and Cladding, which is Insensitive to bending,Easy peeling,Easy welding,ensures small optical loss and…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Planned Expansion and Performance Monitoring
Amazon plans to expand RNG deployments to additional data centers worldwide throughout 2026, with continuous monitoring and optimization. Further research aims to refine routing algorithms, improve cabling methods, and develop tools for easier deployment and diagnostics. Industry adoption and potential standardization are also anticipated.
Klein Tools VDV526-200 Cable Tester, LAN Scout Jr. 2 Ethernet Tester for CAT 5e, CAT 6/6A Cables with RJ45 Connections
VERSATILE CABLE TESTING: Cable tester for data (RJ45) terminated cables and patch cords, ensuring comprehensive testing capabilities
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is RNG in data center networks?
RNG, or Resilient Network Graphs, is a flat, random graph-based topology designed to improve resilience, efficiency, and scalability in data center networks by replacing traditional hierarchical architectures.
How does RNG compare to traditional fat-tree networks?
RNG reduces the number of routers by about 69%, delivers higher throughput (33% more), cuts power consumption by 40%, and lowers operating costs by 27%, while offering more predictable resilience.
Are these RNG networks ready for widespread adoption?
Initial deployments are operational and demonstrate promising results, but further large-scale validation and industry testing are needed before broader adoption can be confirmed.
What challenges remain for RNG deployment?
Long-term operational stability, integration with existing infrastructure, and scaling beyond current deployments are ongoing challenges that Amazon and researchers are actively addressing.
Will RNG influence industry standards?
While promising, RNG’s influence on industry standards will depend on continued validation, performance, and acceptance by other major cloud providers and networking organizations.
Source: Hacker News
