面向算力互联的快速光交换技术研究（特邀）

尹林; 王富; 王小龙; 郭文魁; 宗一宸

doi:10.13756/j.gtxyj.2024.240032

您当前的位置：

首页 >

文章列表页 >

面向算力互联的快速光交换技术研究（特邀）

专题：数据中心内光交换 | 更新时间：2024-10-08

- 面向算力互联的快速光交换技术研究（特邀）
- Fast Optical Switching for Computing-Power Interconnects
- 光通信研究 2024年第5期页码：24003201-24003208
- 作者机构：
  
  1.北京邮电大学　信息光子学与光通信全国重点实验室，北京　100876
  2.齐鲁工业大学（山东省科学院）山东省计算中心（国家超级计算济南中心）算力互联网与信息安全教育部重点实验室，济南　250014
  3.山东省计算机网络重点实验室　山东省基础科学研究中心（计算机科学），济南　250014
- 作者简介：
  
  尹林（2001－），男，贵州遵义人。硕士，主要研究方向为光交换技术。
  王富，副研究员。E-mail：wangfu@bupt.edu.cn
- 基金信息：
  
  国家自然科学基金资助项目(62206017);算力互联网与信息安全教育部重点实验室开放课题资助项目(2023ZD001)
- DOI：10.13756/j.gtxyj.2024.240032
  中图分类号： TN929
- 纸质出版日期：2024-10-10，
  
  收稿日期：2024-02-08，
  
  修回日期：2024-04-08，
- 稿件说明：
移动端阅览
尹林，王富，王小龙，等. 面向算力互联的快速光交换技术研究[J]. 光通信研究，2024(5)：240032.

Yin L, Wang F, Wang X L，et al. Fast Optical Switching for Computing-Power Interconnects[J]. Study on Optical Communications，2024(5)：240032.
尹林，王富，王小龙，等. 面向算力互联的快速光交换技术研究[J]. 光通信研究，2024(5)：240032. DOI： 10.13756/j.gtxyj.2024.240032.

Yin L, Wang F, Wang X L，et al. Fast Optical Switching for Computing-Power Interconnects[J]. Study on Optical Communications，2024(5)：240032. DOI： 10.13756/j.gtxyj.2024.240032.

摘要

随着大模型等人工智能（AI）技术的发展，以及“东数西算”等国家重大工程的深入推进，智算中心（AIDC）和算力网络（CPNs）的内部网络互联成为光网络发展的重要目标。相较于传统面向数据的信息互联，AIDC和CPNs对传输的时延、吞吐量和确定性具有更高的需求。一方面，AIDC内部需要实现高密度和低时延的数据处理和计算资源调度，以满足大规模实时训练和推理；另一方面，AIDC之间互联要求高带宽、低时延和高可靠性的数据传输及资源共享，以支持跨中心的数据协同和灾备恢复。因此，算力互联（CPI）网络需要光交换技术来支撑AIDC内部传送网络的高效数据传送和AIDC之间的带宽快速调度。构建面向CPI的光交换技术将服务于CPI需求，为算力服务提供高效、确定和安全的数据交互管道。文章从网络架构、数据交换和应用场景等方面对光交换技术进行了分析。

Abstract

With the development of Artificial Intelligence(AI)technologies such as large language models and the promotion of major national projects such as "Naional Somputing Network to Synergize East and West"

the internal network interconnection of Artificial Intelligence Data Centers (AIDC) and Computing-Power Networks (CPNs) has become an important tendency of optical network development. Compared with the traditional data-oriented interconnection

the AIDC networks and CPNs have a higher demand for transmission delay

throughput and determinism. On the one hand

high density

low-latency processing and computing resource scheduling are required within the AIDC to meet large-scale real-time training and reasoning. On the other hand

interconnection between the AIDCs requires high-bandwidth

low-latency

high-reliability data transmission and resource sharing to support multi-data-center collaboration and failure recovery. Therefore

the Computing-Power Interconnection (CPI) needs optical switching technology to support efficient data transmission in the optical transport network within the AIDCs and fast bandwidth scheduling between the AIDCs. Hence

constructing a CPI-oriental optical switching network can accommodate the AIDC network requirements and provide efficient

deterministic

and reliable data services. In this paper

we analyze the optical switching technology for AIDC networks

including CPI architecture

data switching and application scenarios.

关键词

智算中心算力互联光交换

Keywords

AIDCCPIoptical switching

references

张逸然, 耿慧拯, 粟栗, 等. 算力网络业务安全技术研究[J]. 移动通信, 2022, 46(11): 90-96.

Zhang Y R, Geng H Z, Su L, et al. Research on Security Technology for Computing Force Network Service[J]. Mobile Communications, 2022, 46(11): 90-96.

马思聪, 孙吉斌, 孙一豪. 东数西算场景下的算力网关研发及应用[J]. 中兴通讯技术, 2023, 29(4): 2-7.

Ma S C, Sun J B, Sun Y H. Research and Application of Computing Power Gateway in East-data-west-computing Project[J]. ZTE Technology Journal, 2023, 29(4): 2-7.

中华人民共和国国家发展和改革委员会. 中华人民共和国国民经济和社会发展第十四个五年规划和2035年远景目标纲要[R]. 北京: 中华人民共和国国家发展和改革委员会, 2021.

National Development and Reform Commission. Outline of the Fourteenth Five-Year Plan for the National Economic and Social Development of the People’s Republic of China and Vision 2035[R]. Beijing: National Development and Reform Commission, 2021.

浪潮信息. 2023-2024年中国人工智能计算力发展评估报告[R]. 济南: 浪潮信息, 2023.

Wave Information. 2023-2024 China Artificial Intelligence Computing Power Development Assessment Report [R]. Jinan: Wave Information, 2023.

张东旭. 面向云服务的光互联网络技术研究[D]. 北京: 北京邮电大学, 2015.

Zhang D X. Study of Cloud Service Oriented Optical Interconnection Network Technologies[D]. Beijing, China: Beijing University of Posts and Telecommunications, 2015.

Choi M C, Lee S, Roh S, et al. A 2.5-32 Gb/s Gen 5-PCIe Receiver with Multi-rate CDR Engine and Hybrid DFE[J]. IEEE Transactions on Circuits and Systems II: Express Briefs, 2022, 69(6): 2677-2681.

Gui T, Wang X, Tang M, et al. Real-time Demonstration of Homodyne Coherent Bidirectional Transmission for Next-generation Data Center Interconnects[J]. Journal of Lightwave Technology, 2021, 39(4): 1231-1238.

Xie C, Zhang B. Scaling Optical Interconnects for Hyperscale Data Center Networks[J]. Proceedings of the IEEE, 2022, 110(11): 1699-1713.

中兴通讯. 中国移动携手中兴通讯完成国内首个集中式算力路由方案试点[EB/OL]. (2022-12-28) [2024-02-09].https://www.zte.com.cn/china/about/news/20221228c2.htmlhttps://www.zte.com.cn/china/about/news/20221228c2.html.

ZTE Corporation. China Mobile and ZTE Complete China’s First Centralized Computing-Power Routing Scheme Pilot [EB/OL]. (2022-12-28) [2024-02-09].https://www.zte.com.cn/china/about/news/20221228c2.htmlhttps://www.zte.com.cn/china/about/news/20221228c2.html.

王光全, 满祥锟, 徐博华, 等. 确定性光传输支撑广域长距算力互联[J]. 邮电设计技术, 2024(2): 7-13.

Wang G Q, Man X K, Xu B H, et al. Deterministic Optical Transmission for Wide Area and Long-distance Computing Power Interconnection[J]. Designing Techniques of Posts and Telecommunications, 2024(2): 7-13.

黄光平, 罗鉴, 周建锋. 算力网络架构与场景分析[J]. 信息通信技术, 2020, 14(4): 16-22.

Huang G P, Luo J, Zhou J F. Analysis of Computation Network Architecture and According Scenarios[J]. Information and Communications Technologies, 2020, 14(4): 16-22.

IEEE Std 802.1Qcc-2018, IEEE Standard for Local and Metropolitan Area Networks: Bridges and Bridged Networks: Amendment 31: Stream Reservation Protocol (SRP) Enhancements and Performance Improvements[S].

中国电子技术标准化研究院. 时间敏感网络白皮书[EB/OL]. (2020-11-25)[2024-02-09].https://www.cesi.cn/images/editor/20201125/20201125082908324001.pdfhttps://www.cesi.cn/images/editor/20201125/20201125082908324001.pdf.

China Electronics Standardization Institute. Time-Sensitive Networking White Paper [EB/OL]. (2020-11-25)[2024-02-09].https://www.cesi.cn/images/editor/20201125/20201125082908324001.pdfhttps://www.cesi.cn/images/editor/20201125/20201125082908324001.pdf.

孙国玮. 确定性算力网络中的资源调度和路由问题研究[D]. 北京: 北京邮电大学, 2023.

Sun G W. Study on Resourse Scheduling and Routing Problems in Deterministic Computing First Network[D]. Beijing, China: Beijing University of Posts and Telecommunications, 2023.

IEEE Std 802.1Qbv-2015, IEEE Standard for Local and Metropolitan Area Networks: Bridges and Bridged Networks - Amendment 25: Enhancements for Scheduled Traffic[S].

池颖英, 朱海龙, 李庆, 等. 基于TSN的智能变电站流量确定性传输[J]. 北京邮电大学学报, 2024, 47(2): 110-117.

Chi Y Y, Zhu H L, Li Q, et al. Deterministic Transmission Of Intelligent Substation Traffic based on TSN[J]. Journal of Beijing University of Posts and Telecommunications, 2024, 47(2): 110-117.

曾鹏程. 时间确定性网络多路径路由算法与协议研究[D]. 西安: 西安电子科技大学, 2021.

Zeng P C. Research on Multipath Routing Algorithms and Protocols in Time Deterministic Networks[D]. Xi’an, China: Xidian University, 2021.

叶沁丹, 范贵生, 黄衡阳. 算力网络一体化支撑方案及应用场景探索[J]. 数据与计算发展前沿, 2022, 4(6): 55-66.

Ye Q D, Fan G S, Huang H Y. Study on Integrated Support Solution and Application Scenarios of Computing Power Network[J]. Frontiers of Data & Computing, 2022, 4(6): 55-66.

王甫涵, 郝祥勇, 蔡轶, 等. 全光数据中心互联的混合放大技术研究[J]. 光通信研究, 2023(3): 6-9.

Wang F H, Hao X Y, Cai Y, et al. Research on Hybrid Amplification Technology of All-optical Data Center Interconnection[J]. Study on Optical Communications, 2023(3): 6-9.

Shi W, Cao J, Zhang Q, et al. Edge Computing: Vision and Challenges[J]. IEEE Internet of Things Journal, 2016, 3(5): 637-646.

Yan F, Miao W, Raz O, et al. Opsquare: a Flat DCN Architecture based on Flow-controlled Optical Packet Switches[J]. Journal of Optical Communications and Networking, 2017, 9(4): 291-303.

Xie C. Optical Interconnects for Data Centers[C]//2016 IEEE Photonics Conference (IPC). Waikoloa, HI, USA: IEEE, 2016: 7830977.

浏览量

下载量

CSCD

CNKI被引量

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于以太无损网络的智算中心光网络架构研究（特邀）

无光缓存多波长光交换的阻塞性能分析

光突发交换技术

DWDM全光网络及光纤光栅在其中的应用