download0 view125
twitter facebook

공공누리This item is licensed Korea Open Government License

Title
Towards effective scheduling policies for many-task applications Practice and experience based on HTCaaS
Author(s)
김직수황순욱Bui QuangVincent Breton김상완김서영노승우
Publication Year
2017-11-10
Abstract
Inthispaper,weconductacomparativestudyofrelativelysimpleyeteffectiveschedulingpoli-ciesformany-taskapplicationswheremultipleuserswithvaryingnumbersoftasksareactivelysharingacommonsysteminfrastructure.Wehaveimplementedthreedifferentschedulingmech-anismsthatcanaddressfairnessanduserresponsetimerespectivelyinacommonmiddlewarestackcalledHTCaaS,whichisapilot-job–basedmulti-levelschedulingsystemrunningontop ofproduction-levelclusters.Asarepresentativecaseofourmany-taskapplications,wehaveleveragedthevirtualscreeningapplication,whichisacomputationaltechniqueusedindrugdis-coveryprocesstoselectthemostpromisingcandidatedrugsforinvitrotestingfrommillionsofchemicalcompounds,whichtypicallyrequiresasubstantialamountofcomputingresources andefficientprocessingofdockingsimulations.Ourcomparativeexperimentalresultsofdiffer-entschedulingpoliciesshowhowwecaneffectivelysupportmultipleusersinasharedresource environmentbybalancingbetweenusersatisfactionandoverallsystemperformance,provideguidelinestoimprovesystemutilization,andaddressadditionaltechnicalchallengestosupport variousmany-task applications.
Keyword
HTCaaS; many-taskcomputing; multi-levelscheduling; resourceallocation; virtualscreening
Journal Title
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
Citation Volume
29
ISSN
1532-0626
Files in This Item:
There are no files associated with this item.
Appears in Collections:
7. KISTI 연구성과 > 학술지 발표논문
URI
https://repository.kisti.re.kr/handle/10580/14670
Export
RIS (EndNote)
XLS (Excel)
XML

Browse