TL;DR
Get home appliances delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
Amazon has successfully built and deployed large-scale data centers using RNG—flat, random graph-based networks—achieving significant efficiency and resilience improvements. This marks a shift away from traditional hierarchical architectures.
Amazon has announced the deployment of flat, random graph-based data center networks, called Resilient Network Graphs (RNG), at scale across its global infrastructure. This development follows years of research and small-scale testing, and it signifies a major shift in data center networking, emphasizing simplicity, resilience, and efficiency.
In 2024, Amazon built its first RNG data center near Dublin, Ireland, after extensive simulation and testing. The RNG architecture replaces traditional hierarchical fat-tree networks with a flat, randomized topology that leverages advanced routing schemes and passive optical cabling. By 2025, Amazon expanded this design to data centers in Germany and Spain, demonstrating its scalability and operational viability.
Compared to conventional fat-tree networks, RNG-based data centers use 69% fewer routers, deliver 33% higher throughput, reduce network power consumption by 40%, and lower operating costs by 27%, according to internal reports. The design achieves these efficiencies through innovative routing algorithms like Spraypoint and cabling solutions such as ShuffleBox, which facilitate high performance and resilience. Amazon officials confirmed that RNG is now the default design for most new data centers globally.
Transforming Data Center Networking with RNG
This shift to RNG networks at Amazon demonstrates a significant advancement in data center architecture, reducing hardware and energy costs while increasing reliability and scalability. It challenges traditional hierarchical designs, offering a model that can be more resilient to failures and easier to operate at large scale, potentially influencing industry standards.
passive optical cabling for data centers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From Hierarchies to Random Graphs in Data Centers
Research into optimal network topologies dates back to the 1970s, with mathematicians exploring expanders—highly connected graphs that prevent isolation of network segments. The networking industry traditionally adopted hierarchical topologies like fat-trees, which, while scalable, are complex to cable and less resilient to failures. In recent years, concepts like Jellyfish and randomized load balancing have challenged this paradigm, but large-scale deployment remained elusive due to routing and operational complexities.
In 2023, AWS researchers began investigating flat, random network designs inspired by mathematical models such as Penrose tiling. After extensive simulation and development of new algorithms and cabling techniques, they successfully demonstrated the feasibility of RNG at scale, culminating in actual data center deployments in 2024 and 2025.
“This is a game-changer—building RNG-based data centers has allowed us to cut costs, reduce power, and improve performance significantly.”
— Matt Rehder, VP of Network Engineering, Amazon
high-performance network switches for data centers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Unresolved Challenges and Performance Guarantees
While initial results are promising, it remains unclear how RNG networks will perform at even larger scales or in diverse operational conditions. Long-term resilience, maintenance complexity, and integration with existing infrastructure are still being evaluated. Industry adoption will depend on further validation and standardization of these methods.
network cabling solutions for large-scale data centers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Scaling RNG and Industry Adoption Pathways
Amazon plans to expand RNG deployments to more data centers worldwide in 2026, refining the design based on operational data. Industry analysts expect other cloud providers and data center operators to monitor Amazon’s results closely, with potential for broader adoption if scalability and reliability are confirmed. Research into standardization and optimization continues within AWS and academia.
resilient data center networking equipment
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is RNG in data center networking?
RNG, or Resilient Network Graphs, are flat, randomized network topologies that use mathematical principles to improve resilience, efficiency, and scalability in data centers.
How does RNG compare to traditional fat-tree networks?
Compared to fat-trees, RNG networks use fewer routers, consume less power, and provide higher throughput, while maintaining or improving resilience to failures.
Is RNG deployment limited to Amazon?
Currently, RNG is an Amazon-specific design, but the underlying research and promising results may influence broader industry adoption in the future.
What are the main technical innovations behind RNG?
Key innovations include the Spraypoint routing scheme, ShuffleBox optical cabling, and software tools for design validation and deployment, all built to work with existing hardware.
What challenges remain before RNG becomes industry standard?
Further validation at larger scales, operational testing in diverse environments, and industry-wide standardization are needed before RNG can be widely adopted beyond Amazon.
Source: Hacker News
NFL season / tailgating Picks
team gear
As an affiliate, we earn on qualifying purchases.
