Stochastic modeling and optimization of stragglers

Farshid Farhat, Diman Zad Tootaghaj, Yuxiong He, Anand Sivasubramaniam, Mahmut Kandemir, Chita R. Das

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

MapReduce framework is widely used to parallelize batch jobs since it exploits a high degree of multi-tasking to process them. However, it has been observed that when the number of servers increases, the map phase can take much longer than expected. This paper analytically shows that the stochastic behavior of the servers has a negative effect on the completion time of a MapReduce job, and continuously increasing the number of servers without accurate scheduling can degrade the overall performance. We analytically model the map phase in terms of hardware, system, and application parameters to capture the effects of stragglers on the performance. Mean sojourn time (MST), the time needed to sync the completed tasks at a reducer, is introduced as a performance metric and mathematically formulated. Following that, we stochastically investigate the optimal task scheduling which leads to an equilibrium property in a datacenter with different types of servers. Our experimental results show the performance of the different types of schedulers targeting MapReduce applications. We also show that, in the case of mixed deterministic and stochastic schedulers, there is an optimal scheduler that can always achieve the lowest MST.

Original languageEnglish (US)
Article number7450631
Pages (from-to)1164-1177
Number of pages14
JournalIEEE Transactions on Cloud Computing
Volume6
Issue number4
DOIs
StatePublished - Oct 1 2018

Fingerprint

Servers
Scheduling
Multitasking
Hardware

All Science Journal Classification (ASJC) codes

  • Software
  • Information Systems
  • Hardware and Architecture
  • Computer Science Applications
  • Computer Networks and Communications

Cite this

Farhat, Farshid ; Tootaghaj, Diman Zad ; He, Yuxiong ; Sivasubramaniam, Anand ; Kandemir, Mahmut ; Das, Chita R. / Stochastic modeling and optimization of stragglers. In: IEEE Transactions on Cloud Computing. 2018 ; Vol. 6, No. 4. pp. 1164-1177.
@article{b6a82a83e471435a80ba43c2b0d17d04,
title = "Stochastic modeling and optimization of stragglers",
abstract = "MapReduce framework is widely used to parallelize batch jobs since it exploits a high degree of multi-tasking to process them. However, it has been observed that when the number of servers increases, the map phase can take much longer than expected. This paper analytically shows that the stochastic behavior of the servers has a negative effect on the completion time of a MapReduce job, and continuously increasing the number of servers without accurate scheduling can degrade the overall performance. We analytically model the map phase in terms of hardware, system, and application parameters to capture the effects of stragglers on the performance. Mean sojourn time (MST), the time needed to sync the completed tasks at a reducer, is introduced as a performance metric and mathematically formulated. Following that, we stochastically investigate the optimal task scheduling which leads to an equilibrium property in a datacenter with different types of servers. Our experimental results show the performance of the different types of schedulers targeting MapReduce applications. We also show that, in the case of mixed deterministic and stochastic schedulers, there is an optimal scheduler that can always achieve the lowest MST.",
author = "Farshid Farhat and Tootaghaj, {Diman Zad} and Yuxiong He and Anand Sivasubramaniam and Mahmut Kandemir and Das, {Chita R.}",
year = "2018",
month = "10",
day = "1",
doi = "10.1109/TCC.2016.2552516",
language = "English (US)",
volume = "6",
pages = "1164--1177",
journal = "IEEE Transactions on Cloud Computing",
issn = "2168-7161",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "4",

}

Stochastic modeling and optimization of stragglers. / Farhat, Farshid; Tootaghaj, Diman Zad; He, Yuxiong; Sivasubramaniam, Anand; Kandemir, Mahmut; Das, Chita R.

In: IEEE Transactions on Cloud Computing, Vol. 6, No. 4, 7450631, 01.10.2018, p. 1164-1177.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Stochastic modeling and optimization of stragglers

AU - Farhat, Farshid

AU - Tootaghaj, Diman Zad

AU - He, Yuxiong

AU - Sivasubramaniam, Anand

AU - Kandemir, Mahmut

AU - Das, Chita R.

PY - 2018/10/1

Y1 - 2018/10/1

N2 - MapReduce framework is widely used to parallelize batch jobs since it exploits a high degree of multi-tasking to process them. However, it has been observed that when the number of servers increases, the map phase can take much longer than expected. This paper analytically shows that the stochastic behavior of the servers has a negative effect on the completion time of a MapReduce job, and continuously increasing the number of servers without accurate scheduling can degrade the overall performance. We analytically model the map phase in terms of hardware, system, and application parameters to capture the effects of stragglers on the performance. Mean sojourn time (MST), the time needed to sync the completed tasks at a reducer, is introduced as a performance metric and mathematically formulated. Following that, we stochastically investigate the optimal task scheduling which leads to an equilibrium property in a datacenter with different types of servers. Our experimental results show the performance of the different types of schedulers targeting MapReduce applications. We also show that, in the case of mixed deterministic and stochastic schedulers, there is an optimal scheduler that can always achieve the lowest MST.

AB - MapReduce framework is widely used to parallelize batch jobs since it exploits a high degree of multi-tasking to process them. However, it has been observed that when the number of servers increases, the map phase can take much longer than expected. This paper analytically shows that the stochastic behavior of the servers has a negative effect on the completion time of a MapReduce job, and continuously increasing the number of servers without accurate scheduling can degrade the overall performance. We analytically model the map phase in terms of hardware, system, and application parameters to capture the effects of stragglers on the performance. Mean sojourn time (MST), the time needed to sync the completed tasks at a reducer, is introduced as a performance metric and mathematically formulated. Following that, we stochastically investigate the optimal task scheduling which leads to an equilibrium property in a datacenter with different types of servers. Our experimental results show the performance of the different types of schedulers targeting MapReduce applications. We also show that, in the case of mixed deterministic and stochastic schedulers, there is an optimal scheduler that can always achieve the lowest MST.

UR - http://www.scopus.com/inward/record.url?scp=85058264535&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85058264535&partnerID=8YFLogxK

U2 - 10.1109/TCC.2016.2552516

DO - 10.1109/TCC.2016.2552516

M3 - Article

AN - SCOPUS:85058264535

VL - 6

SP - 1164

EP - 1177

JO - IEEE Transactions on Cloud Computing

JF - IEEE Transactions on Cloud Computing

SN - 2168-7161

IS - 4

M1 - 7450631

ER -