网络计算技术发展情况及应用前景分析

doi:10.12267/j.issn.2096-5931.2023.06.007

摘要/Abstract

摘要：

随着网络速度持续提升,网络处理开销愈发显著,网络计算技术已成为卸载主机负担、释放服务器算力、解决网络瓶颈的关键技术,在云计算和边缘计算领域备受关注。首先从网络计算技术的需求入手,阐述网络计算技术产生的背景,重点围绕可编程交换和智能网卡两方面,梳理网络计算技术的发展情况和产业化情况,最终对网络计算技术的应用前景进行分析,给出其应用场景和应用价值。

关键词: 网络计算, 智能网卡, 数据处理单元, 可编程交换

Abstract:

With the continuous improvement of network speeds, the network processing overhead becomes increasingly significant. The in-network computing technology has become a key technology to offload the burden of hosts, release computing power of server, and solve network bottlenecks, which has attracted much attention in the fields of cloud computing and edge computing. First, starting with the needs of in-network computing, the background of in-network computing is described. Then, the development and industrialization of in-network computing technology is sorted out by focusing on the two aspects, namely, programmable switch and smart Network Interface Card (NIC). Finally, the application prospect of in-network computing is analyzed, and the application scenarios and the application value are proposed.

Key words: in-network computing, smart NIC, Data Processing Unit, programmable switch

中图分类号:

TP393.1

王翰华, 李峰. 网络计算技术发展情况及应用前景分析[J]. 信息通信技术与政策, 2023, 49(6): 47-54.

WANG Hanhua, LI Feng. Analysis of development and application prospect of in-network computing technology[J]. Information and Communications Technology and Policy, 2023, 49(6): 47-54.

导出引用管理器 EndNote|Ris|BibTeX

链接本文:

http://ictp.caict.ac.cn/CN/10.12267/j.issn.2096-5931.2023.06.007

http://ictp.caict.ac.cn/CN/Y2023/V49/I6/47

图/表 3

参考文献 38

[1]	THOMAS S, VOELKER G M, PORTER G. CacheCloud: towards speed-of-light datacenter communication[C]//Proceedings of the 10th USENIX Conference on Hot Topics in Cloud Computing. Boston, MA, USA: USENIX Association, 2018: 7.
[2]	班有容, 王锦涛, 王鹏, 等. DPU技术在算力网络中的思考与实践[J]. 通信世界, 2022(17): 46-49.
[3]	KIANPISHEH S, TALEB T. A survey on in-network computing: programmable data plane and technology specific applications[J]. IEEE Communications Surveys & Tutorials, 2023, 25(1): 701-761.
[4]	PORTS D R K, NELSON J. When should the network be the computer?[C/OL]//Proceedings of the Workshop on Hot Topics in Operating Systems. New York, NY, USA: Association for Computing Machinery, 2019: 209-215[2022-05-31].
[5]	TOKUSASHI Y, MATSUTANI H, ZILBERMAN N. LaKe: the power of in-network computing[C/OL]//2018 International Conference on ReConFigurable Computing and FPGAs(ReConFig). Cancun, Mexico: IEEE, 2018: 1-8[2023-03-26]. https://ieeexplore.ieee.org/document/8641696/
[6]	NUNES B A A, MENDONCA M, NGUYEN X N, et al. A survey of software-defined networking: past, present, and future of programmable networks[J]. IEEE Communications Surveys & Tutorials, 2014, 16(3): 1617-1634.
[7]	MCKEOWN N, ANDERSON T, BALAKRISHNAN H, et al. OpenFlow: enabling innovation in campus networks[J]. ACM SIGCOMM Computer Communication Review, 2008, 38(2): 69-74.
[8]	BIFULCO R, RETVARI G. A survey on the programmable data plane: abstractions, architectures, and open problems[C/OL]//2018 IEEE 19th International Conference on High Performance Switching and Routing (HPSR). Bucharest, Romania: IEEE, 2018: 1-7[2023-04-24]. https://ieeexplore.ieee.org/document/8850761/
[9]	SANKARAN G C, SIVALINGAM K M, GONDALIYA H. P4 and NetFPGA-based secure in-network computing architecture for AI-enabled industrial internet of things[J/OL]. IEEE Internet of Things Journal, 2023, 10(4): 2979-2994. https://ieeexplore.ieee.org/abstract/document/9606196 doi: 10.1109/JIOT.2021.3125862 URL
[10]	PFAFF B, PETTIT J, KOPONEN T, et al. The design and implementation of Open vSwitch[C]//NSDI’15: Proceedings of the 12th USENIX Conference on Networked Systems Design and Implementation. Berkeley: USENIX Association, 2015: 117-130.
[11]	DPDK Project. Programmer’s guide:data plane development kit 23.03.0 documentation[EB/OL]. (2023-03-31)[2023-04-24]. https://doc.dpdk.org/guides/prog_guide/
[12]	张昕怡, 潘恒, 谢高岗. 可编程网络数据平面技术进展[J]. 电信科学, 2022, 38(6): 42-50. doi: 10.11959/j.issn.1000-0801.2022131
[13]	HAN S, JANG S, CHOI H, et al. Virtualization in programmable data plane: a survey and open challenges[J]. IEEE Open Journal of the Communications Society, 2020(1): 527-534.
[14]	BAS A. Leveraging stratum and tofino fast refresh for software upgrades[EB/OL]. (2018-12-06)[2023-04-24]. https://opennetworking.org/wp-content/uploads/2018/12/Tofino_Fast_Refresh.pdf
[15]	YAN Y, BELDACHI A F, NEJABATI R, et al. P4-enabled smart NIC: enabling sliceable and service-driven optical data centres[J]. Journal of Lightwave Technology, 2020, 38(9): 2688-2694. doi: 10.1109/JLT.50 URL
[16]	CHEN R, SUN G. A survey of kernel-bypass techniques in network stack[C/OL]//Proceedings of the 2018 2nd International Conference on Computer Science and Artificial Intelligence. Shenzhen, China: ACM, 2018: 474-477[2023-04-25]. https://dl.acm.org/doi/10.1145/3297156.3297242
[17]	QIU Y, XING J, HSU K F, et al. Automated smartNIC offloading insights for network functions[C/OL]//Proceedings of the ACM SIGOPS 28th Symposium on Operating Systems Principles(SOSP’21). New York, NY, USA: ACM, 2021: 772-787[2023-04-25]. https://dl.acm.org/doi/10.1145/3477132.3483583
[18]	OSINSKI T, KOSSAKOWSKI M, TARASIUK H, et al. Offloading data plane functions to the multi-tenant cloud infrastructure using P4[C/OL]// 2019 ACM/IEEE Symposium on Architectures for Networking and Communications Systems (ANCS). Cambridge, United Kingdom: IEEE, 2019: 1-6[2023-04-25].https://ieeexplore.ieee.org/document/8901895/
[19]	刘忠沛, 吕高锋, 王继昌, 等. 专用数据处理器综述[J]. 计算机工程与科学, 2023, 45(2): 215-227.
[20]	BARSELLOTTI L, ALHAMED F, VEGAS OLMOS J J, et al. Introducing data processing units (DPU) at the edge [Invited][C/OL]//2022 International Conference on Computer Communications and Networks (ICCCN). Honolulu, HI, USA: IEEE, 2022: 1-6[2023-03-27]. https://ieeexplore.ieee.org/document/9868927/
[21]	BOSSHART P, DALY D, GIBB G, et al. P4: programming protocol-independent packet processors[J]. ACM SIGCOMM Computer Communication Review, 2014, 44(3): 87-95.
[22]	HAUSER F, HÄBERLE M, MERLING D, et al. A survey on data plane programming with P4: Fundamentals, advances, and applied research[J]. Journal of Network and Computer Applications, 2023, 212: 103561. doi: 10.1016/j.jnca.2022.103561 URL
[23]	KHOOI X Z, CSIKOR L, LI J, et al. In-network applications: beyond single switch pipelines[C/OL]//2021 IEEE 7th International Conference on Network Softwarization (NetSoft). Tokyo, Japan: IEEE, 2021: 1-8[2023-04-25]. https://ieeexplore.ieee.org/document/9492665/
[24]	Intel. Intel^© Tofino^TM 3 intelligent fabric processors[EB/OL]. (2021-10-27)[2023-04-23]. https://www.intel.cn/content/www/cn/zh/products/network-io/programmable-ethernet-switch/tofino-3-brief.html
[25]	Marvell. Marvell^© Teralynx^© 10 data center ethernet switch[EB/OL]. (2023-03-23)[2023-04-23]. https://www.marvell.com/content/dam/marvell/en/public-collateral/switching/marvell-teralynx-10-data-center-ethernet-switch-product-brief.pdf
[26]	Cisco. Solutions - Cisco silicon one G100 data sheet[EB/OL]. (2021-10-25)[2023-04-26]. https://www.cisco.com/c/en/us/solutions/collateral/silicon-one/datasheet-c78-744833.html
[27]	ARCILLA A. Broadcom trident 3 platform performance analysis[EB/OL]. 2019[2023-04-23]. https://docs.broadcom.com/doc/12395356
[28]	NPL org. GitHub - nplang/NPL-Spec[EB/OL]. (2019-07-11)[2023-04-26]. https://github.com/nplang/NPL-Spec
[29]	PUTNAM A, CAULFIELD A M, CHUNG E S, et al. A reconfigurable fabric for accelerating large-scale datacenter services[J]. ACM SIGARCH Computer Architecture News, 2014, 42(3): 13-24. doi: 10.1145/2678373.2665678 URL
[30]	CAULFIELD A, COSTA P, GHOBADI M. Beyond SmartNICs: towards a fully programmable cloud: invited paper[C/OL]//2018 IEEE 19th International Conference on High Performance Switching and Routing (HPSR). Bucharest, Romania: IEEE, 2018: 1-6[2023-04-28]. https://ieeexplore.ieee.org/document/8850757
[31]	马潇潇, 杨帆, 王展, 等. 智能网卡综述[J]. 计算机研究与发展, 2022, 59(1): 1-21.
[32]	BURSTEIN I. Nvidia data center processing unit (DPU) architecture[C/OL]//2021 IEEE Hot Chips 33 Symposium (HCS). Palo Alto, CA, USA: IEEE, 2021: 1-20[2023-03-27]. https://ieeexplore.ieee.org/document/9567066/
[33]	Intel. Intel unveils infrastructure processing unit roadmap; 2nd generation to ship in 2022[EB/OL]. (2022-05-10)[2023-04-23]. https://download.intel.com/newsroom/2022/corporate/vision/Intel-IPU-Roadmap-Fact-Sheet.pdf
[34]	AMD. DSC200 distributed services card[EB/OL]. 2022[2023-04-23]. https://www.amd.com/system/files/documents/pensando-dsc-200-product-brief.pdf
[35]	Broadcom. Broadcom stingray smartNIC accelerates baidu cloud services[EB/OL]. (2020-03-31)[2023-04-28]. https://www.broadcom.com/company/news/product-releases/53106
[36]	Marvell. Marvell^© OCTEON 10 DPU platform product brief[EB/OL]. (2021-06-21)[2023-04-28]. https://www.marvell.com/content/dam/marvell/en/public-collateral/embedded-processors/marvell-octeon-10-dpu-platform-product-brief.pdf
[37]	LIGUORI A. The nitro project: next generation AWS infrastructure[C]//Hot Chips: A Symposium on High Performance Chips. Palo Alto, CA, USA:IEEE, 2018:1-39.
[38]	阿里云开发者社区. 阿里云张献涛:自主最强DPU神龙的秘诀[EB/OL]. (2022-02-07)[2023-04-28]. https://developer.aliyun.com/article/867094

企业名称	产品名称	架构
英伟达	Bluefield-3	SoC
Marvell	OCTEON 10	SoC
英特尔	Mount Evans	SoC
英特尔	F2000X-PL	CPU+FPGA
博通	Stingray	SoC
AMD	Pensando DSC Elba	SoC

企业名称	产品名称	架构
英伟达	Bluefield-3	SoC
Marvell	OCTEON 10	SoC
英特尔	Mount Evans	SoC
英特尔	F2000X-PL	CPU+FPGA
博通	Stingray	SoC
AMD	Pensando DSC Elba	SoC

企业名称	产品类型		产品名称	备注
英伟达	智能网卡	ASIC	ConnectX	收购Mellanox
英伟达	智能网卡	DPU SoC	Bluefield	收购Mellanox
Marvell	智能网卡	DPU SoC	OCTEON
Marvell	可编程交换芯片		Teralynx
英特尔	智能网卡	DPU SoC	Mount Evans
	智能网卡	CPU+FPGA	F2000X-PL	收购Altera
	可编程交换芯片		Tofino	收购Barefoot
博通	智能网卡	DPU SoC	Stingray
博通	可编程交换芯片		Trident
AMD	智能网卡	DPU	DSC	收购Pensando
AMD	智能网卡	FPGA	Alveo	收购赛灵思

企业名称	产品类型		产品名称	备注
英伟达	智能网卡	ASIC	ConnectX	收购Mellanox
英伟达	智能网卡	DPU SoC	Bluefield	收购Mellanox
Marvell	智能网卡	DPU SoC	OCTEON
Marvell	可编程交换芯片		Teralynx
英特尔	智能网卡	DPU SoC	Mount Evans
	智能网卡	CPU+FPGA	F2000X-PL	收购Altera
	可编程交换芯片		Tofino	收购Barefoot
博通	智能网卡	DPU SoC	Stingray
博通	可编程交换芯片		Trident
AMD	智能网卡	DPU	DSC	收购Pensando
AMD	智能网卡	FPGA	Alveo	收购赛灵思