D 2016

Complex Job Scheduling Simulations with Alea 4

KLUSÁČEK, Dalibor, Gabriela PODOLNÍKOVÁ a Šimon TÓTH

Základní údaje

Originální název

Complex Job Scheduling Simulations with Alea 4

Autoři

KLUSÁČEK, Dalibor (203 Česká republika, garant), Gabriela PODOLNÍKOVÁ (203 Česká republika, domácí) a Šimon TÓTH (203 Česká republika)

Vydání

Belgium, Proceedings of the 9th EAI International Conference on Simulation Tools and Techniques (SimuTools), od s. 124-129, 6 s. 2016

Nakladatel

ICST

Další údaje

Jazyk

angličtina

Typ výsledku

Stať ve sborníku

Obor

10201 Computer sciences, information science, bioinformatics

Stát vydavatele

Belgie

Utajení

není předmětem státního či obchodního tajemství

Forma vydání

elektronická verze "online"

Kód RIV

RIV/00216224:14330/16:00114973

Organizační jednotka

Fakulta informatiky

ISBN

978-1-63190-120-1

Klíčová slova anglicky

Alea; simulation; job scheduling; workload adaptation

Příznaky

Mezinárodní význam, Recenzováno
Změněno: 31. 3. 2021 15:33, RNDr. Pavel Šmerk, Ph.D.

Anotace

V originále

Simulations have been used for many years in order to either develop, refine or validate new setups of production resource managers, that are used in large computing systems such as HPC clusters, computing grids or clouds. Typically, new job scheduling algorithms or various resource-related policies are first evaluated in a simulator prior to their deployment. To facilitate this, accurate models of both the applied resource manager and the workload being processed are very important to obtain reliable simulation outputs. In this paper we present a new major release of the Alea simulator, that has been developed in order to allow for such detailed and realistic simulations. The simulator allows for detailed emulation of typical scheduling systems that are widely used in nowadays computing centers. Furthermore, it also provides novel approach to properly model dynamic user-to-system interactions. We also present recent real-life based examples where the Alea simulator has been used to improve the performance of an actual computing system, demonstrating its practical capabilities and usefulness.