arrow
返回

Minimizing Buffer Utilization for Lossless Inter-DC Links

delete2024-12-01
delete0
PRE
AI
C
Chengyuan Huang
F
Feiyang Xue
P
Peiwen Yu
X
Xiaoliang Wang
Y
Yanqing Chen
T
Tao Wu
L
Lei Han
Z
Zifa Han
B
Bingquan Wang
X
Xiangyu Gong
C
Chen Tian *
W
Wanchun Dou
G
Guihai Chen
H
Hao Yin
DOI:10.1109/TNET.2024.3443600delete
delete原文链接
delete原文求助
delete分享
delete收藏
摘要

摘要

En 中文
RDMA over Converged Ethernet (RoCEv2) has been widely deployed to data centers (DCs) for its better compatibility with Ethernet/IP than Infiniband (IB). As cross-DC applications emerge, they also demand high throughput, low latency, and lossless network for cross-DC data transmission. However, RoCEv2's underlying lossless mechanism Priority-based Flow Control (PFC) cannot fit into the long-haul transmission scenario and degrades the performance of RoCEv2. PFC is myopic and only considers queue length to pause upstream senders, which leads to large queueing delay. This paper proposes Bifrost, a downstream-driven lossless flow control that supports long distance cross-DC data transmission. Bifrost uses virtual incoming packets, which indicates the upper bound of in-flight packets, together with buffered packets to control the flow rate. It minimizes the buffer space requirement to one-hop bandwidth delay product (BDP) and achieves low one-way latency. Moreover, we extend Bifrost and propose BifrostX, to accommodate the multi-priority queue of the current switch implementation. BifrostX enables flow control for each queue separately while maintaining low buffer reservation, no throughput loss, and no packet loss. Real-world experiments are conducted with prototype switches and 80 kilometers cables. Evaluations demonstrate that compared to PFC, Bifrost reduces average/tail flow completion time (FCT) of inter-DC flows by up to 22.5%/42.0%, respectively. Bifrost is compatible with existing infrastructure and can support distance of thousands of kilometers.
Keyword:
Data center
RoCEv2
flow control
cross-DC data transmission

期刊

I
IEEE-ACM Transactions on Networking
IF:
3.6
论文数:
4.4K
被引数:
9.5K

机构

H
huawei technologies
学者数:
3.3K
论文数: 2.9K
被引数: 1
T
tsinghua university
学者数:
11.9W
论文数: 10.0W
被引数: 137
N
nanjing university
学者数:
7.8W
论文数: 5.6W
被引数: 87
学者 查看更多机构
引用论文

引用论文

Swing: Providing Long-Range Lossless RDMA via PFC-RelaySwing: 通过PFC继电器提供长距离无损RDMA
err2023-01-01
err3
PREAI
errChen, Yanqing; Tian, Chen; Dong, Jiaqing; Feng, Song; Zhang, Xu; Liu, Chang; Yu, Peiwen; Xia, Nai; Dou, Wanchun; Chen, Guihai
err分享
err收藏
Struktur und Synthese eines spirocyclischen Phosphoranid‐Anions: Modell für einen Zwischenzustand bei einer S N 2(P)‐Reaktion
err2006-01-23
err0
PREAI
errDietmar Schomburg; Werner Strozer; Rainer Bohlen; Winfried Kuhn; Gerd‐Volker Röschenthaler
err分享
err收藏
Catalytic enantioselective protonation of samarium enolates by a C2-symmetric chiral diol
err1996-04-01
err0
PREAI
errYutaka Nakamura; Seiji Takeuchi; Akiko Ohira; Yoshiaki Ohgo
err分享
err收藏
Studies of Potassium Tungsten Bronzes Containing Lithium or Sodium.
err1987-01-01
err0
errOAAI
errAltaf Hussain; Lars Kihlborg; Knut Maartmann-Moe; Otto Maberg; Alf Scheie; D. Louër
err分享
err收藏
Intra-session test-retest reliability of functional connectivity in infants
err2021-10-01
err0
errOAAI
errYun Wang; Walter Hinds; Cristiane S Duarte; Seonjoo Lee; Catherine Monk; Melanie Wall; Glorisa Canino; Ana Carolina C. Milani; Andrea Jackowski; Marina Griorgi Mamin; Bernd U. Foerster; Jay Gingrich; Myrna M Weissman; Bradley S. Peterson; David Semanek; Edna Acosta Perez; Eduardo Labat; Ioannisely Berrios Torres; Ivaldo Da Silva; Camila Parente; Nitamar Abdala; Jonathan Posner
err分享
err收藏
Case series of diffuse extraneural metastasis in H3F3A mutant high-grade gliomas: Clinical, molecular phenotype and literature review
err2021-07-01
err0
PREAI
errSana Mohiuddin; Stefania Maraka; Muhammad Usman Baig; Sumit Gupta; Tariq Muzzafar; Tibor Valyi-Nagy; Holly Lindsay; Karen Moody; Shehla Razvi; Arnold Paulino; Konstantin Slavin; Vinai Gondi; Ian McCutcheon; Wafik Zaky; Soumen Khatua
err分享
err收藏
err分享
err收藏
学者 查看更多内容