Apache Nemo: A Framework for Optimizing Distributed Data Processing

Author:

Song Won Wook1,Yang Youngseok1,Eo Jeongyoon1,Seo Jangho2,Kim Joo Yeon3,Lee Sanha2,Lee Gyewon1,Um Taegeon1,Cho Haeyoon1,Chun Byung-Gon1

Affiliation:

1. Seoul National University, Seoul, Rep. of Korea

2. Naver Corporation, Gyeonggi-do, Rep. of Korea

3. Samsung Electronics, Seoul, Rep. of Korea

Abstract

Optimizing scheduling and communication of distributed data processing for resource and data characteristics is crucial for achieving high performance. Existing approaches to such optimizations largely fall into two categories. First, distributed runtimes provide low-level policy interfaces to apply the optimizations, but do not ensure the maintenance of correct application semantics and thus often require significant effort to use. Second, policy interfaces that extend a high-level application programming model ensure correctness, but do not provide sufficient fine control. We describe Apache Nemo, an optimization framework for distributed dataflow processing that provides fine control for high performance and also ensures correctness for ease of use. We combine several techniques to achieve this, including an intermediate representation of dataflow, compiler optimization passes, and runtime extensions. Our evaluation results show that Nemo enables composable and reusable optimizations that bring performance improvements on par with existing specialized runtimes tailored for a specific deployment scenario. Apache Nemo is open-sourced at https://nemo.apache.org as an Apache incubator project.

Funder

Institute of Information & Communications Technology Planning & Evaluation

Korea government

BK21 FOUR Intelligence Computing

National Research Foundation of Korea

Publisher

Association for Computing Machinery (ACM)

Subject

General Computer Science

Reference57 articles.

1. Bert Hubert. 2020. Linux Traffic Control. Retrieved from https://lartc.org/manpages/tc.txt.

2. CAIDA. 2020. The CAIDA Anonymized Internet Traces 2016 Dataset. Retrieved from https://www.caida.org/data/passive/passive_2016_dataset.xml.

Cited by 2 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. Blaze: Holistic Caching for Iterative Data Processing;Proceedings of the Nineteenth European Conference on Computer Systems;2024-04-22

2. SWAN;Proceedings of the 13th ACM SIGOPS Asia-Pacific Workshop on Systems;2022-08-23

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3