UB-Mesh: A Hierarchically Localized nD-FullMesh Data Center Network Architecture

The scaling of Large-scale Language Models (LLMs) demands unprecedented computational power and bandwidth. We present UB-Mesh, an innovative AI datacenter network architecture that enhances scalability, performance, and cost-efficiency through a hierarchical nD-FullMesh topology. Unlike traditional symmetrical designs, UB-Meshoptimizes LLM training by prioritizing localized data movement and minimizing switch usage. The architecture features UB-Mesh-Pod, a physical implementation of 4D-FullMesh using custom hardware including NPUs, CPUs, Low/High-Radix Switches (LRS/HRS), and NICs, interconnected via our Unified Bus (UB) technology for dynamic resource allocation. For network optimization, we introduce All-Path-Routing (APR) to efficiently manage data traffic. Combined with topology-aware performance tuning and robust reliability mechanisms like 64 + 1 backup, UB-Meshachieves 2.04× better cost-efficiency and 7.2% higher availability than Clos networks. These innovations address the critical challenges of building practical, high-performance AI infrastructure at scale.

Authors

Institutions

Publication Details

Journal
IEEE Micro
Published
2025-09-01
DOI
https://doi.org/10.1109/mm.2025.3592688
Citations
8
Primary Topic
Cloud Computing and Resource Management
Type
article
Field-Weighted Citation Impact
25.97
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

UB-Mesh: A Hierarchically Localized nD-FullMesh Data Center Network Architecture

Wenhua Du, Zhe Zhou, Yi Xiong, Xianping Chen et al.
8 citations
IEEE Micro
Cloud Computing and Resource Management
25.97
article

UB-Mesh: A Hierarchically Localized nD-FullMesh Data Center Network Architecture

Wenhua Du, Zhe Zhou, Yi Xiong, Xianping Chen, Rui Meng, Yan‐Zhi Gai, Zhongjun Cheng, Bingyang Liu, Xiangyu Chen, Yu Zhou, Heng Liao, C. C. Li, Xingyue Li, Pei-Chen Wu, Chen Lv, Jing Xia, Jintian Gao, Yong Qin, Yongsheng Ma, Bill McColl, Xi Chen, Lei Chen, Ziyang Zhang, Christopher H.K. Cheng, Wenjie Liu, Jian He, Buyun Wang, Peng Dong, Shengyuan Yan, Jianbing Wang, Chuan Qian, Zhigang Guo, Li Xiong, Shizhong Xiao
article en
8 citations

Abstract

The scaling of Large-scale Language Models (LLMs) demands unprecedented computational power and bandwidth. We present UB-Mesh, an innovative AI datacenter network architecture that enhances scalability, performance, and cost-efficiency through a hierarchical nD-FullMesh topology. Unlike traditional symmetrical designs, UB-Meshoptimizes LLM training by prioritizing localized data movement and minimizing switch usage. The architecture features UB-Mesh-Pod, a physical implementation of 4D-FullMesh using custom hardware including NPUs, CPUs, Low/High-Radix Switches (LRS/HRS), and NICs, interconnected via our Unified Bus (UB) technology for dynamic resource allocation. For network optimization, we introduce All-Path-Routing (APR) to efficiently manage data traffic. Combined with topology-aware performance tuning and robust reliability mechanisms like 64 + 1 backup, UB-Meshachieves 2.04× better cost-efficiency and 7.2% higher availability than Clos networks. These innovations address the critical challenges of building practical, high-performance AI infrastructure at scale.

IEEE MicroVol. 45(5)
Huawei Technologies (China) (CN)
Openalex Percentile: Top 1%
Cloud Computing and Resource Management
25.97
DOIOpen Access PDFOpenAlex
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.