Presentation 2013-01-25
Enhancing Hadoop Performance by Network-Aware Task Scheduling
Shun KATAOKA, Hyuma WATANABE, Masatoshi KAWARASAKI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Hadoop uses MapReduce for distributed processing of large-scale data. Although a large amount of data transfer between Mappers and Reducers may be generated in a shuffle phase of the MapReduce processing, current Hadoop task-scheduling method allocates a task without taking this into account. As a result, job processing may be delayed because of the network bottleneck. In this paper, we investigate Hadoop data flow in detail and propose a new task-scheduling method that allocates a new task based on the observed amount of data transfer at each node. Job performance in the proposed task-scheduling method is evaluated in the real machine environment.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Hadoop / MapReduce / task scheduling / network / job performance
Paper # IN2012-152
Date of Issue

Conference Information
Committee IN
Conference Date 2013/1/17(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Information Networks (IN)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Enhancing Hadoop Performance by Network-Aware Task Scheduling
Sub Title (in English)
Keyword(1) Hadoop
Keyword(2) MapReduce
Keyword(3) task scheduling
Keyword(4) network
Keyword(5) job performance
1st Author's Name Shun KATAOKA
1st Author's Affiliation College of Media Arts, Science and Technology, University of Tsukuba()
2nd Author's Name Hyuma WATANABE
2nd Author's Affiliation Graduate School of Library, Information and Media Studies, University of Tsukuba
3rd Author's Name Masatoshi KAWARASAKI
3rd Author's Affiliation Graduate School of Library, Information and Media Studies, University of Tsukuba
Date 2013-01-25
Paper # IN2012-152
Volume (vol) vol.112
Number (no) 393
Page pp.pp.-
#Pages 6
Date of Issue