RNG: Flat Datacenter Networks at Scale
The paper presents RNG, a new design for flat datacenter networks that utilizes quasi-random graphs. RNG features a distributed routing protocol that enhances fault tolerance and reduces costs compared to traditional network topologies. It has been adopted as the default network for most workloads at Amazon, demonstrating superior performance and cost efficiency.
- ▪RNG is based on quasi-random graphs and aims to improve datacenter network design.
- ▪The new distributed routing protocol allows for multiple edge disjoint paths between endpoints.
- ▪RNG is up to 45% cheaper than traditional fat tree networks and matches or exceeds their performance.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Networking and Internet Architecture arXiv:2604.15261 (cs) [Submitted on 16 Apr 2026 (v1), last revised 21 May 2026 (this version, v3)] Title:RNG: Flat Datacenter Networks at Scale Authors:Giacomo Bernardi, Ratul Mahajan, C. Seshadhri, Enrico Carlesso, Chinchu Merine Joseph, Saurabh Kumar, Pavan Manikonda, Luiza Popa, Randy Ram, Steven Robinson, Elizabeth Tennent View a PDF of the paper titled RNG: Flat Datacenter Networks at Scale, by Giacomo Bernardi and 10 other authors View PDF HTML (experimental) Abstract:We design and deploy in production the first flat datacenter networks. Our design, called RNG, is based on quasi-random graphs.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv.org.
Discussion
0 commentsMore from arXiv.org
