Skip to main navigation Skip to search Skip to main content

RPC: Joint Online Reducer Placement and Coflow Bandwidth Scheduling for Clusters

  • SUNY Buffalo
  • Nanjing University

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

10 Scopus citations

Abstract

Reducing Coflow Completion Time (CCT) has a significant impact on application performance in data-parallel frameworks. Most existing works assume that the endpoints of constituent flows in each coflow are predetermined. We argue that CCT can be further optimized by treating flows' destinations as an additional optimization dimension via reducer placement. In this paper, we propose and implement RPC, a joint online Reducer Placement and Coflow bandwidth scheduling framework, to minimize the average CCT in cloud clusters. We first develop a 2-approximation algorithm to minimize the CCT of a single coflow, then schedule all the coflows following the Shortest Remaining Time First (SRTF) principle. We use a real testbed implementation and extensive large-scale simulations to demonstrate that RPC can reduce the average CCT by 64.98% compared with state-of-the-art technologies.

Original languageEnglish
Title of host publicationProceedings - 26th IEEE International Conference on Network Protocols, ICNP 2018
PublisherIEEE Computer Society
Pages187-197
Number of pages11
ISBN (Electronic)9781538660430
DOIs
StatePublished - Nov 7 2018
Event26th IEEE International Conference on Network Protocols, ICNP 2018 - Cambridge, United Kingdom
Duration: Sep 24 2018Sep 27 2018

Publication series

NameProceedings - International Conference on Network Protocols, ICNP
Volume2018-September
ISSN (Print)1092-1648

Conference

Conference26th IEEE International Conference on Network Protocols, ICNP 2018
Country/TerritoryUnited Kingdom
CityCambridge
Period09/24/1809/27/18

Keywords

  • Cloud computing
  • Flow scheduling
  • Reducer placement

Fingerprint

Dive into the research topics of 'RPC: Joint Online Reducer Placement and Coflow Bandwidth Scheduling for Clusters'. Together they form a unique fingerprint.

Cite this