AutoSketch: Automatic Sketch-Oriented Compiler for Query-driven Network Telemetry

Authors: 

Haifeng Sun and Qun Huang, National Key Laboratory for Multimedia Information Processing, School of Computer Science, Peking University; Jinbo Sun, Institute of Computing Technology, Chinese Academy of Sciences; Wei Wang, Northeastern University, China; Jiaheng Li, National Key Laboratory for Multimedia Information Processing, School of Computer Science, Peking University; Fuliang Li, Northeastern University, China; Yungang Bao, Institute of Computing Technology, Chinese Academy of Sciences; Xin Yao and Gong Zhang, Huawei Theory Department

Abstract: 

Recent network telemetry witnesses tremendous progress in two directions: query-driven telemetry that targets expressiveness as the primary goal, and sketch-based algorithms that address resource-accuracy trade-offs. In this paper, we propose AutoSketch that aims to integrate the advantages of both classes. In a nutshell, AutoSketch automatically compiles high-level operators into sketch instances that can be readily deployed with low resource usage and incur limited accuracy drop. However, there remains a gap between the expressiveness of high-level operators and the underlying realization of sketch algorithms. AutoSketch bridges this gap in three aspects. First, AutoSketch extends its interface derived from existing query-driven telemetry such that users can specify the desired telemetry accuracy. The specified accuracy intent will be utilized to guide the compiling procedure. Second, AutoSketch leverages various techniques, such as syntax analysis and performance estimation, to construct efficient sketch instances. Finally, AutoSketch automatically searches for the most suitable parameter configurations that fulfill the accuracy intent with minimum resource usage. Our experiments demonstrate that AutoSketch can achieve high expressiveness, high accuracy, and low resource usage compared to state-of-the-art telemetry solutions.

NSDI '24 Open Access Sponsored by
King Abdullah University of Science and Technology (KAUST)

Open Access Media

USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.

BibTeX
@inproceedings {295665,
author = {Haifeng Sun and Qun Huang and Jinbo Sun and Wei Wang and Jiaheng Li and Fuliang Li and Yungang Bao and Xin Yao and Gong Zhang},
title = {{AutoSketch}: Automatic {Sketch-Oriented} Compiler for Query-driven Network Telemetry},
booktitle = {21st USENIX Symposium on Networked Systems Design and Implementation (NSDI 24)},
year = {2024},
isbn = {978-1-939133-39-7},
address = {Santa Clara, CA},
pages = {1551--1572},
url = {https://www.usenix.org/conference/nsdi24/presentation/sun},
publisher = {USENIX Association},
month = apr
}