設計與實作基於OVS-DPDK並結合硬體加速的適應性流量卸載策略

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：101

、訪客IP：3.148.106.49

姓名

李兆偉(Chao-Wei Lee) 查詢紙本館藏

畢業系所

資訊管理學系

論文名稱

設計與實作基於OVS-DPDK並結合硬體加速的適應性流量卸載策略
(Design and Implementation of an Adaptive Flow Offloading Strategy with Hardware-Accelerated OVS-DPDK)

相關論文

★ 設計與實作基於軟體定義網路之 SRv6 適應性路徑選擇方法

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

至系統瀏覽論文 (2027-9-1以後開放)

摘要(中)

現今雲端資料中心結合軟體定義網路 (Software-Defined Networking, SDN)、網路功能虛擬化 (Network Function Virtualization, NFV) 與硬體卸載 (Hardware Offloading) 等技術，能夠以可程式化網路方式來強化資料中心的網路架構和加速封包處理過程，使網路基礎設施能承載更多的流量、以及具備更穩定的運作狀態。Open vSwitch (OVS) 是一種SDN/NFV架構下常見的虛擬交換器，能利用智慧型網路卡 (SmartNIC) 將轉發規則卸載至硬體轉發表，進一步加速封包轉發的效率。然而硬體轉發表空間有限，需建立合適的卸載策略以充分利用硬體轉發表的空間。然而一般研究主要是基於固定轉發規則卸載數量上限的同時，盡可能地將所有合適的轉發規則進行卸載與替換，因此無法根據當前網路流量的變化，動態調整所需使用的硬體轉發表空間。
本研究將以資料中心網路的運作情境為背景，結合網路虛擬化架構的軟硬體元件，提出一流量感知的適應性流量卸載 (TA-AFO) 策略。TA-AFO透過持續監控OVS-DPDK上的流量變化，動態調整觸發轉發規則卸載到SmartNIC的門檻值，以減少硬體空間的消耗並確保轉發效率。除此之外，亦將TA-AFO整合至OVS-DPDK中，並使用Nvidia BlueField2智慧型200G網路卡進行驗證。實驗結果顯示，在搭配真實網路流量的模擬環境中，TA-AFO與TFO及Elixir相較之下，平均轉發速率最高可分別提升約18% 及10%，而Drop Rate則可降低至94% 及 93%，並且能夠更快地對網路流量的變化進行反應。

摘要(英)

Nowadays, cloud data centers integrate technologies such as Software-Defined Networking (SDN), Network Function Virtualization (NFV), and Hardware Offloading to strengthen the network architecture and accelerate packet processing through programmable networking, allowing the network infrastructure to handle more traffic and maintain more stable operations. Open vSwitch (OVS) is a common virtual switch in SDN/NFV architectures that utilizes SmartNICs to offload forwarding rules to hardware forwarding tables, significantly accelerating packet forwarding efficiency. However, the limited space in hardware forwarding tables necessitates the development of suitable offloading strategies to maximize their utilization. Most research typically focuses on offloading and replacing all suitable forwarding rules up to a fixed limit, without the capability to dynamically adjust the required hardware forwarding table space based on current network traffic changes.
This study leverages the operational scenarios of data center networks to propose a Traffic-Aware Adaptive Flow Offloading (TA-AFO) strategy, integrating the software and hardware components of network virtualization architectures. TA-AFO continuously monitors traffic changes on OVS-DPDK and dynamically adjusts the thresholds for offloading forwarding rules to SmartNICs, thereby reducing hardware space consumption and ensuring forwarding efficiency. Furthermore, TA-AFO will be integrated into OVS-DPDK and validated using 200G Nvidia BlueField-2 SmartNICs. Experimental results show that, in a simulated environment with real network traffic, TA-AFO achieves an average forwarding rate improvement of up to approximately 18% and 10% compared to TFO and Elixir, respectively. Additionally, the Drop Rate is reduced by up to 94% and 93%, and TA-AFO responds more quickly to changes in network traffic.

關鍵字(中)

★ 軟體定義網路
★ 封包處理
★ 硬體卸載
★ 智慧型網路卡
★ Open vSwitch

關鍵字(英)

★ SDN
★ Packet Processing
★ Hardware Offloading
★ SmartNIC
★ Open vSwitch

論文目次

摘要 I
Abstract II
目錄 III
圖目錄 V
表目錄 VI
一、緒論 1
1.1. 研究背景 1
1.2. 研究動機 2
1.3. 研究目的及預期貢獻 2
1.4. 章節架構 4
二、背景知識與相關研究 5
2.1. Software-Defined Networking 5
2.2. Data Plane Development Kit 6
2.3. Open vSwitch 8
2.4. SmartNIC 10
2.5. 討論 12
三、系統設計 13
3.1. 系統架構 13
3.2. 資料層 14
3.2.1. dpif-netdev模組 14
3.2.2. ofproto-dpif-upcall模組 15
3.3. 控制層 15
3.3.1. 轉發規則的統計資訊蒐集之運作流程 15
3.3.2. TA-AFO模組 16
3.3.3. stats-collection模組 20
四、實驗與評估 22
4.1. 實驗環境 22
4.2. 實驗結果與討論 25
4.2.1. OVS-DPDK已匹配flow卸載之功能驗證 25
4.2.2. SmartNIC卸載流量增加對OVS-DPDK的影響 26
4.2.3. 卸載策略評估 27
4.3. 實驗總結 33
五、結論與未來研究 34
參考文獻 35

參考文獻

[1] Worldwide Semiannual Public Cloud Services Tracker. [Online]. Available: https://www.idc.com/getdoc.jsp?containerId=IDC_P29737 (Accessed: 7 July 2024)
[2] D. Kreutz, et al., “Software-Defined Networking: A Comprehensive Survey,” in proceedings of the IEEE, vol. 103, no. 1, pp. 14-76, 2015.
[3] D. Firestone, et al., “Azure accelerated networking: SmartNICs in the public cloud,” in proceedings of the USENIX Conference on Networked Systems Design and Implementation, pp. 51-66, 2018.
[4] S. Jain, et al., “B4: experience with a globally-deployed software defined wan,” in proceedings of the ACM SIGCOMM Computer Communication Review, vol. 43, pp 3-14, 2013.
[5] R. Mijumbi, et al., “Network Function Virtualization: State-of-the-Art and Research Challenges,” in proceedings of the IEEE Communications Surveys and Tutorials, vol. 18, no. 1, pp. 236-262, 2016.
[6] M. Wang, et al., “Service Function Chain Composition and Mapping in NFV-Enabled Networks,” in proceedings of the IEEE World Congress on Services, pp. 331-334, 2019.
[7] B. Hardin, D. Comer and A. Rastegarnia, “DCnet: Evaluation Of A New Data Center Network Architecture,” in proceedings of the Conference on Innovation in Clouds, Internet and Networks and Workshops, pp. 11-15, 2023.
[8] B. Pfaff, et al., “The design and implementation of open vSwitch,” in proceedings of the USENIX Conference on Networked Systems Design and Implementation, pp. 117-130, 2015.
[9] W. Tu, et al., “Revisiting the open vSwitch dataplane ten years later,” in proceedings of the ACM SIGCOMM Conference, pp. 245-257, 2021.
[10] Q. Cai, et al., “Understanding host network stack overheads,” in proceedings of the ACM SIGCOMM Conference, pp. 65-77, 2021.
[11] Intel DPDK: Data plane development kit. [Online]. Available: https://www.dpdk.org/ (Accessed: 7 July 2024)
[12] Update to VMware’s per-CPU Pricing Model VMware. [Online]. Available: https://news.vmware.com/company/cpu-pricing-model-update-feb-2020 (Accessed: 7 July 2024)
[13] Mellanox ASAP2 Accelerated Switching and Packet Processing. [Online]. Available: https://network.nvidia.com/files/doc-2020/sb-asap2.pdf (Accessed: 7 July 2024)
[14] TruFlow. [Online]. Available: https://techdocs.broadcom.com/us/en/storage-and-ethernet-connectivity/ethernet-nic-controllers/bcm957xxx/adapters/introducing-truflow.html (Accessed: 7 July 2024)
[15] Improving 5G Performance Using OvS Over ASAP² with AMD EPYC 7002 and NVIDIA Mellanox SmartNICs. [Online]. Available: https://developer.nvidia.com/blog/improving-5g-performance-using-ovs-over-asap-with-amd-epyc-7002-and-nvidia-mellanox-smartnics/ (Accessed: 7 July 2024)
[16] P. Gao, Y. Xu, and H. J. Chao, “OVS-CAB: Efficient rule-caching for Open vSwitch hardware offloading,” in proceedings of the Computer Networks, vol. 188, pp. 107844, 2021.
[17] T. Benson, A. Akella, and D. A. Maltz, “Network traffic characteristics of data centers in the wild,” in proceedings of the ACM SIGCOMM Conference on Internet measurement, pp. 267-280, 2010.
[18] F. Yan, et al., “Network Traffic Characteristics of Hyperscale DataCenters in the Era of Cloud Applications,” in proceedings of the Journal of Optical Communications and Networking, vol. 15, no. 10, pp. 736-749, 2023.
[19] N. Sarrar, et al., “Leveraging Zipf′s law for traffic offloading,” in proceedings of the SIGCOMM Computer Communication Review, vol. 42, pp. 16-22, 2012.
[20] R. Durner and W. Kellerer, “Network Function Offloading Through Classification of Elephant Flows,” in proceedings of the IEEE Transactions on Network and Service Management, vol. 17, no. 2, pp. 807-820, 2020.
[21] N. Sarrar, et al., “Elixir: A High-performance and Low-cost Approach to Managing {Hardware/Software} Hybrid Flow Tables Considering Flow Burstiness,” in proceedings of the USENIX Symposium on Networked Systems Design and Implementation, pp. 535-550, 2022.
[22] Y. Zhang, et al., “On the characteristics and origins of internet flow rates,” in proceedings of the Conference on Applications, Technologies, Architectures, and Protocols for Computer Communications, pp. 309-322, 2002.
[23] S. Bauer, et al., “On the evolution of internet flow characteristics,” in proceedings of the Applied Networking Research Workshop, pp. 29-35, 2021.
[24] Y. Li, et al., “Serverless Computing: State-of-the-Art, Challenges and Opportunities,” in proceedings of the IEEE Transactions on Services Computing, vol. 16, no. 2, pp. 1522-1539, 2023.
[25] H. Li, K. Ota and M. Dong, “Virtual Network Recognition and Optimization in SDN-Enabled Cloud Environment,” in proceedings of the IEEE Transactions on Cloud Computing, vol. 9, no. 2, pp. 834-843, 2021.
[26] F. Tang, et al., “Elephant Flow Detection and Load-Balanced Routing with Efficient Sampling and Classification,” in proceedings of the IEEE Transactions on Cloud Computing, vol. 9, no. 3, pp. 1022-1036, 2021.
[27] I. Maity, S. Misra and C. Mandal, “DART: Data Plane Load Reduction for Traffic Flow Migration in SDN” in proceedings of the IEEE Transactions on Communications, vol. 69, no. 3, pp. 1765-1774, 2021.
[28] DPDK Overview. [Online]. Available: https://edc.intel.com/content/www/us/en/design/products/ethernet/config-guide-e810-dpdk/dpdk-overview/ (Accessed: 7 July 2024)
[29] Q. Ren, et al., “PacketUsher: Exploiting DPDK to accelerate compute-intensive packet processing,” in proceedings of the Computer Communications, vol. 161, pp. 323-333, 2020.
[30] Introduction to Hyperscan. [Online]. Available: https://www.intel.com/content/www/us/en/developer/articles/technical/introduction-to-hyperscan.html (Accessed: 7 July 2024)
[31] Multiple vDPI Functions using DPDK and Hyperscan on OVS-DPDK Platform. [Online]. Available: https://www.dpdk.org/wp-content/uploads/sites/35/2018/05/DPDK_Summit18_Multiple-vDPI-Functions-using-DPDK-and-Hyperscan-on-OVS-DPDK-Platform_final.pdf (Accessed: 7 July 2024)
[32] openvswitch. [Online]. Available: https://docs.openvswitch.org/en/latest/ (Accessed: 16 July 2024)
[33] OVS-DPDK Datapath Classifier. [Online]. Available: https://www.intel.com/content/www/us/en/developer/articles/technical/ovs-dpdk-datapath-classifier.html (Accessed: 16 July 2024)
[34] V. Srinivasan, S. Suri and G. Varghese, “Packet classification using tuple space search,” in proceedings of the Conference on Applications, Technologies, Architectures, and Protocols for Computer Communication, vol. 29, pp. 135-146, 1999.
[35] OpenFlow Switch Specification. [Online]. Available: https://opennetworking.org/wp-content/uploads/2014/10/openflow-switch-v1.5.1.pdf (Accessed: 7 July 2024)
[36] J. L. Hennessy and D. A. Patterson, “A new golden age for computer architecture,” in proceedings of the Communications of the ACM, vol. 62, no. 2, pp. 48-60, 2019.
[37] Xilinx Alveo U25 SmartNIC Platform Launch. [Online]. Available: https://www.xilinx.com/publications/product-briefs/xilinx-smartNIC-media-briefing-final.pdf (Accessed: 7 July 2024)
[38] T. Cui, et al., “Offloading load balancers onto SmartNICs,” in proceedings of the ACM SIGOPS Asia-Pacific Workshop on Systems, pp. 56-62, 2021.
[39] S. Sun, et al., “SKV: A SmartNIC-Offloaded Distributed Key-Value Store,” in proceedings of the IEEE International Conference on Cluster Computing, pp. 1-11, 2022.
[40] OpenVswitch hardware offload over DPDK. [Online]. Available: https://fast.dpdk.org/events/slides/DPDK-2017-11-OVS_Hardware_Offload.pdf (Accessed: 7 July 2024)
[41] NVIDIA BlueField Networking Platform. [Online]. Available: https://www.nvidia.com/en-us/networking/products/data-processing-unit/ (Accessed: 19 July 2024)
[42] Intel Infrastructure Processing Unit. [Online]. Available: https://www.intel.com/content/www/us/en/products/details/network-io/ipu.html (Accessed: 19 July 2024)
[43] AMD Pensando DPU Accelerators. [Online]. Available: https://www.amd.com/en/products/accelerators/pensando.html (Accessed: 19 July 2024)
[44] Intel FPGA SmartNIC. [Online]. Available: https://www.intel.com/content/www/us/en/products/details/fpga/platforms/smartnic.html (Accessed: 19 July 2024)
[45] AMD Alveo Adaptable Accelerator Cards. [Online]. Available: https://www.amd.com/en/products/accelerators/alveo.html (Accessed: 19 July 2024)
[46] W. Yang and C. Chang, “SmartGate: Accelerate Cloud Gateway with SmartNIC,” in proceedings of the International Conference on Cyber Security and Information Engineering, pp. 8-14, 2023.
[47] Intelligent Traffic Offload Uses SmartNIC/DPU for Hyperscale Security. [Online]. Available: https://www.paloaltonetworks.com/blog/2021/07/intelligent-traffic-offload/ (Accessed: 7 July 2024)
[48] Y. Dong, et al., “High performance network virtualization with SR-IOV,” in proceedings of the International Symposium on High-Performance Computer Architecture, pp. 1-10, 2010.
[49] TRex. [Online]. Available: https://trex-tgn.cisco.com/ (Accessed: 7 July 2024)
[50] RFC 2544. [Online]. Available: https://www.ietf.org/rfc/rfc2544.txt (Accessed: 7 July 2024)
[51] G. Cormode and S. Muthukrishnan, “An improved data stream summary: the count-min sketch and its applications,” in proceedings of the Journal of Algorithms, vol. 55, no. 1, pp. 58-75, 2005.
[52] Data Set for IMC 2010 Data Center Measurement. [Online]. Available: https://pages.cs.wisc.edu/~tbenson/IMC10_Data.html (Accessed: 7 July 2024)
[53] htop. [Online]. Available: https://github.com/htop-dev/htop (Accessed: 7 July 2024)

指導教授

蔡邦維陳以錚(Pang-Wei Tsai Yi-Cheng Chen)

審核日期

2024-7-29

推文