Unleashing SmartNIC Packet Processing Performance in P4

被引：8

作者：

Xing, Jiarong ^{[1
]}

Qiu, Yiming ^{[1
]}

Hsu, Kuo-Feng ^{[2
]}

Sui, Songyuan ^{[1
]}

Manaa, Khalid ^{[3
]}

Shabtai, Omer ^{[3
]}

Piasetzky, Yonatan ^{[3
]}

Kadosh, Matty ^{[3
]}

Krishnamurthy, Arvind ^{[4
]}

Ng, T. S. Eugene ^{[1
]}

Chen, Ang ^{[1
]}

机构：

[1] Rice Univ, Houston, TX 77251 USA

[2] Meta, Menlo Pk, CA USA

[3] Nvidia, Santa Clara, CA USA

[4] Univ Washington, Seattle, WA 98195 USA

来源：

PROCEEDINGS OF THE 2023 ACM SIGCOMM 2023 CONFERENCE, SIGCOMM 2023 | 2023年

关键词：

SmartNICs; P4; Runtime Program Optimization;

D O I：

10.1145/3603269.3604882

中图分类号：

TP301 [理论、方法];

学科分类号：

081202 ;

摘要：

SmartNICs are on the rise as a packet processing platform, with the trend towards a uniform P4 programming model. However, unleashing SmartNIC packet processing performance in P4 is a formidable task. Traditional SmartNIC optimizations rely on low-level program tuning, but P4 abstractions operate at one level above. At the same time, today's P4 optimizations primarily focus on resource packing rather than performance tuning. We develop Pipeleon, an automated performance optimization framework for P4 programmable SmartNICs. We introduce techniques that are tailored to the performance characteristics of SmartNICs, and further leverage dynamic workload patterns for profile-guided optimization. Pipeleon pinpoints program hotspots at the P4 level and computes runtime optimization plans to specialize the program layout based on the latest profile. We have prototyped Pipeleon and applied it to optimize two popular P4 SmartNICs-Nvidia BlueField2 and Netronome Agilio CX-as well as a software SmartNIC emulator extended based on BMv2. Our results show that Pipeleon significantly improves SmartNIC packet processing performance in realistic scenarios.

引用

页码：1028 / 1042

页数：15

共 50 条

[31] P4 = innovation
Jones, Samuel M.
ANNALS OF FAMILY MEDICINE, 2007, 5 (03) : 280 - 281
[32] P4: Portable Parallel Processing Pipelines for Interactive Information Visualization
Li, Jianping Kelvin
Ma, Kwan-Liu
IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS, 2020, 26 (03) : 1548 - 1561
[33] Inversion by P4: polarization-picture post-processing
Schechner, Yoav Y.
PHILOSOPHICAL TRANSACTIONS OF THE ROYAL SOCIETY B-BIOLOGICAL SCIENCES, 2011, 366 (1565) : 638 - 648
[34] Design and implementation of P4 virtual switches and P4 virtual networks
Chan, Kwan-Yee
Lee, Steven S. W.
COMPUTER COMMUNICATIONS, 2023, 199 (126-138) : 126 - 138
[35] 买颗P4玩超频——P4超频性能分析与珍品级P4选购
电脑自做, 2004, (12) : 101 - 104
[36] A Survey on P4 Challenges in Software Defined Networks: P4 Programming
Goswami, Bhargavi
Kulkarni, Manasa
Paulose, Joy
IEEE ACCESS, 2023, 11 : 54373 - 54387
[37] On Ramsey (P4, P4) -minimal graphs for small-order
Rahmadani, Desi
Nusantara, Toto
3RD INTERNATIONAL CONFERENCE ON MATHEMATICS AND SCIENCE EDUCATION (ICOMSE) 2019: STRENGTHENING MATHEMATICS AND SCIENCE EDUCATION RESEARCH FOR THE CHALLENGE OF GLOBAL SOCIETY, 2020, 2215
[38] 解析P4未来市场选择P4最佳应用
农业网络信息, 2001, (12) : 35 - 35
[39] Performance evaluation of microPET P4 for rat, rabbit and monkey
Wada, Y
Matsumura, A
Nakamura, F
Tanaka, M
Mizuma, H
Mizuno, K
Nozaki, S
Nakajima, H
Kondo, S
Ishlii, K
Koyama, K
Hamazawa, Y
Okuma, T
Okamura, T
Inoue, Y
Watanabe, Y
QUANTITATION IN BIOMEDICAL IMAGING WITH PET AND MRI, 2004, (1265): : 69 - 73
[40] Model Checking-Based Performance Prediction for P4
Lukacs, Daniel
Pongracz, Gergely
Tejfel, Mate
ELECTRONICS, 2022, 11 (14)

← 1 2 3 4 5 →