An investigation into the impacts of task-level behavioural heterogeneity upon energy efficiency in Cloud datacentres.
Name:
Publisher version
View Source
Access full-text PDFOpen Access
View Source
Check access options
Check access options
Name:
An Investigation into the Impacts ...
Size:
1.141Mb
Format:
PDF
Description:
Author Accepted Manuscript
Abstract
Cloud datacentre resources and the arriving jobs are addressed to be exhibiting increased level of heterogeneity. A single Cloud job may encompass one to several number of tasks, such tasks usually exhibit increased level of behavioural heterogeneity though they belong to the same job. Such behavioural heterogeneity are usually evident among the level of resource consumption, resource intensiveness, task duration etc. These task behavioural heterogeneity within jobs impose various complications in achieving an effective energy efficient management of the Cloud jobs whilst processing them in the server resources. To this end, this paper investigates the impacts of the task level behavioural heterogeneity upon energy efficiency whilst the tasks within given jobs are executed in Cloud datacentres. Real-life Cloud trace logs have been investigated to exhibit the impacts of task heterogeneity from three different perspectives including the task execution trend and task termination pattern, the presence of few proportions of resource intensive and long running tasks within jobs. Furthermore, the energy implications of such straggling tasks within jobs have been empirically exhibited. Analysis conducted in this study demonstrates that Cloud jobs are extremely heterogeneous and tasks behave distinctly under different execution instances, and the presence of energy-aware long tail stragglers within jobs can significantly incur extravagant level of energy expenditures.Citation
Panneerselvam, J. et al (2018) 'An investigation into the impacts of task-level behavioural heterogeneity upon energy efficiency in Cloud datacentres', Future Generation Computer Systems, Volume 83, June 2018, pp. 239-249. DOI: 10.1016/j.future.2017.12.064Publisher
ElsevierJournal
Future Generation Computer SystemsDOI
10.1016/j.future.2017.12.064Additional Links
http://linkinghub.elsevier.com/retrieve/pii/S0167739X1731960XType
ArticleLanguage
enISSN
0167739Xae974a485f413a2113503eed53cd6c53
10.1016/j.future.2017.12.064