Presentation 2018-03-09
Analysis of tasks that continuously generate error events in cloud platforms
Koji Mandai, Jun Kawahara, Shoji Kasahara,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this research, we analyze the characteristics of failed jobs in data sets recording the execution status of jobs and tasks in the data center called Google Cluster-Usage Traces. It is known that a failure job tends to include many tasks that continuously generate error events. In the existing research, methods to reduce the resource usage such as CPU and memory have been proposed. The method interrupts task execution when a certain number of events occur at execution. In this method, we propose an approach to accurately identify unusual tasks by analyzing the causes of events in detail. With this method, we show that it is possible to reduce 6.92¥% of the CPU usage and 3.75¥% of the memory usage among the total resource utilization, and to reduce the risk of interrupting successfully jobs for the analyzed trace data.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Cloud / Job Scheduling / Resource Management
Paper # ICM2017-64
Date of Issue 2018-03-01 (ICM)

Conference Information
Committee ICM
Conference Date 2018/3/8(2days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Toshio Tonouchi(NEC)
Vice Chair Yuji Nomura(Fujitsu Labs.) / Yoichi Yamashita(NTT-N)
Secretary Yuji Nomura(Fujitsu) / Yoichi Yamashita(KDDI R&D Labs.)
Assistant Haruo Ooishi(NTT)

Paper Information
Registration To Technical Committee on Information and Communication Management
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Analysis of tasks that continuously generate error events in cloud platforms
Sub Title (in English)
Keyword(1) Cloud
Keyword(2) Job Scheduling
Keyword(3) Resource Management
1st Author's Name Koji Mandai
1st Author's Affiliation Nara Institute of Science and Technology(NAIST)
2nd Author's Name Jun Kawahara
2nd Author's Affiliation Nara Institute of Science and Technology(NAIST)
3rd Author's Name Shoji Kasahara
3rd Author's Affiliation Nara Institute of Science and Technology(NAIST)
Date 2018-03-09
Paper # ICM2017-64
Volume (vol) vol.117
Number (no) ICM-491
Page pp.pp.49-54(ICM),
#Pages 6
Date of Issue 2018-03-01 (ICM)