Cloud friendly load balancing for HPC applications: Preliminary work

Osman Sarood, Abhishek Gupta, Laxmikant V. Kalé

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Cloud computing is increasingly being explored as a cost effective alternative and addition to supercomputers for some High Performance Computing (HPC) applications. However, dynamic environment and interference by other virtual machines are some of the factors which prevent efficient execution of HPC applications in cloud. Through this research, we leverage a message driven adaptive runtime system to develop techniques that reduce the gap between application performance on cloud and supercomputers. Our scheme uses object migration to achieve load balance for tightly coupled parallel applications executing in virtualized environments that suffer from interfering jobs. While restoring load balance, it not only reduces the timing penalty caused by interfering jobs, but also reduces energy consumption significantly. With experimental evaluation using benchmarks and a real HPC application, we demonstrate that our scheme reduces the timing penalty and energy overhead associated with interfering jobs by at least 50%.

Original languageEnglish (US)
Title of host publicationProceedings - 41st International Conference on Parallel Processing Workshops, ICPPW 2012
Pages200-205
Number of pages6
DOIs
StatePublished - Dec 20 2012
Event41st International Conference on Parallel Processing Workshops, ICPPW 2012 - Pittsburgh, PA, United States
Duration: Sep 10 2012Sep 13 2012

Publication series

NameProceedings of the International Conference on Parallel Processing Workshops
ISSN (Print)1530-2016

Other

Other41st International Conference on Parallel Processing Workshops, ICPPW 2012
CountryUnited States
CityPittsburgh, PA
Period9/10/129/13/12

Keywords

  • Cloud computing
  • energy consumption
  • green clouds
  • load balancing

ASJC Scopus subject areas

  • Software
  • Mathematics(all)
  • Hardware and Architecture

Fingerprint Dive into the research topics of 'Cloud friendly load balancing for HPC applications: Preliminary work'. Together they form a unique fingerprint.

Cite this