With the development of big data, a demand for data analysis keeps increasing. This requirement has prompted a need for data-aware task scheduling approach that can simultaneously schedule various tasks such as batched tasks and real-time tasks in a data center efficiently. To this end, we propose a hybrid task scheduling strategy coupled with data migration in data center. Firstly, we translate the task scheduling problem into task selection problem, and give methods of selecting batched tasks and real-time tasks respectively. Then the method for scheduling both batched tasks and real-time tasks is introduced in detail. Finally, we integrate data migration into the hybrid scheduling strategy. Experimental results show that, compared to the traditional FIFO algorithm, the proposed task scheduling strategy greatly improves the data locality and data migration performs very well on reducing the job execution time. Our algorithm also guarantees an acceptable fairness for tasks.
History
Volume
11335
Pagination
630-644
Location
Guangzhou, China
Start date
2018-11-15
End date
2018-11-17
ISSN
0302-9743
eISSN
1611-3349
ISBN-13
9783030050535
Language
eng
Publication classification
E1 Full written paper - refereed
Copyright notice
2018, Springer Nature Switzerland AG
Editor/Contributor(s)
Vaidya J, Li J
Title of proceedings
ICA3PP 2018 : Proceedings of the International Conference on Algorithms and Architectures for Parallel Processing
Event
Algorithms and Architectures for Parallel Processing. Conference (2018 : Guangzhou, China)
Publisher
Springer
Place of publication
Cham, Switzerland
Series
Algorithms and Architectures for Parallel Processing Conference