Learning scheduling algorithms for data processing clusters
Name
sigcomm-2019.pdf
Description
Accepted version
Size
7.38 MB
Format
Adobe PDF
Checksum (MD5)
c18d74de7c26f8b071f8a5743226748b
Author(s)
Mao, Hongzi
Schwarzkopf, Malte
Venkatakrishnan, Shaileshh Bojja
Meng, Zili
Alizadeh, Mohammad
Date Issued
August 2019
Journal
SIGCOMM 2019 - Proceedings of the 2019 Conference of the ACM Special Interest Group on Data Communication
Publisher
Association for Computing Machinery (ACM)
Citation
Mao, Hongzi, Schwarzkopf, Malte, Venkatakrishnan, Shaileshh Bojja, Meng, Zili and Alizadeh, Mohammad. 2019. "Learning scheduling algorithms for data processing clusters." SIGCOMM 2019 - Proceedings of the 2019 Conference of the ACM Special Interest Group on Data Communication.
Version
Author's final manuscript
Abstract
© 2019 Association for Computing Machinery. Efficiently scheduling data processing jobs on distributed compute clusters requires complex algorithms. Current systems use simple, generalized heuristics and ignore workload characteristics, since developing and tuning a scheduling policy for each workload is infeasible. In this paper, we showthat modern machine learning techniques can generate highly-efficient policies automatically. Decima uses reinforcement learning (RL) and neural networks to learn workload-specific scheduling algorithms without any human instruction beyond a high-level objective, such as minimizing average job completion time. However, off-the-shelf RL techniques cannot handle the complexity and scale of the scheduling problem. To build Decima, we had to develop new representations for jobs' dependency graphs, design scalable RL models, and invent RL training methods for dealing with continuous stochastic job arrivals. Our prototype integration with Spark on a 25-node cluster shows that Decima improves average job completion time by at least 21% over hand-tuned scheduling heuristics, achieving up to 2× improvement during periods of high cluster load.
MIT Department
Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.1145/3341302.3342080
