DLBer: A dynamic load balancing algorithm for the event-driven clusters

  • Mingming Sun
  • , Changlong Li
  • , Xuehai Zhou
  • , Kun Lu
  • , Hang Zhuang

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

Abstract

The event-driven programming model has been proposed to efficiently process iterative applications and incremental applications. In clusters based the event-driven model, applications are structured as a series of triggers, each of which will be invoked when associate events are trigged. And framework assigns a newly submitted trigger to a node where the relevant datasets set. Unfortunately it may lead to load imbalance because associate events occur by chance. Numerous triggers in a node may be simultaneously invoked but other nodes have no triggers running. Jobs composed of short, sub-second triggers present a difficult balancing challenge. To the end, we design DLBer, a new dynamic load balancing algorithm for the event-driven clusters to maximize improve the utilization of node resources.

Original languageEnglish
Title of host publicationNetwork and Parallel Computing - 11th IFIP WG 10.3 International Conference, NPC 2014, Proceedings
PublisherSpringer Verlag
Pages608-611
Number of pages4
ISBN (Print)9783662449165
DOIs
StatePublished - 2014
Externally publishedYes
Event11th IFIP WG 10.3 International Conference on Network and Parallel Computing, NPC 2014 - Ilan, Taiwan, Province of China
Duration: 18 Sep 201420 Sep 2014

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8707 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference11th IFIP WG 10.3 International Conference on Network and Parallel Computing, NPC 2014
Country/TerritoryTaiwan, Province of China
CityIlan
Period18/09/1420/09/14

Fingerprint

Dive into the research topics of 'DLBer: A dynamic load balancing algorithm for the event-driven clusters'. Together they form a unique fingerprint.

Cite this