DeepPlace: Learning to place applications in multi-tenant clusters

Subrata Mitra; Shanka Subhra Mondal; Nikhil Sheoran; Neeraj Dhake; Ravinder Nehra; Ramanuja Simha

doi:10.1145/3343737.3343741

DeepPlace: Learning to place applications in multi-tenant clusters

Subrata Mitra
, Shanka Subhra Mondal
, Nikhil Sheoran
, Neeraj Dhake
, Ravinder Nehra
, Ramanuja Simha

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

5 Scopus citations

Abstract

Large multi-tenant production clusters often have to handle a variety of jobs and applications with a variety of complex resource usage characteristics. It is non-trivial and non-optimal to manually create placement rules for scheduling that would decide which applications should co-locate. In this paper, we present DeepPlace, a scheduler that learns to exploits various temporal resource usage patterns of applications using Deep Reinforcement Learning (Deep RL) to reduce resource competition across jobs running in the same machine while at the same time optimizing for overall cluster utilization.

Original language	English (US)
Title of host publication	Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019
Publisher	Association for Computing Machinery, Inc
Pages	61-68
Number of pages	8
ISBN (Electronic)	9781450368933
DOIs	https://doi.org/10.1145/3343737.3343741
State	Published - Aug 19 2019
Externally published	Yes
Event	10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019 - Hangzhou, China Duration: Aug 19 2019 → Aug 20 2019

Publication series

Name	Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019

Conference

Conference	10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019
Country/Territory	China
City	Hangzhou
Period	8/19/19 → 8/20/19

ASJC Scopus subject areas

Computer Networks and Communications
Hardware and Architecture

Access to Document

10.1145/3343737.3343741

Cite this

Mitra, S., Mondal, S. S., Sheoran, N., Dhake, N., Nehra, R., & Simha, R. (2019). DeepPlace: Learning to place applications in multi-tenant clusters. In Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019 (pp. 61-68). (Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019). Association for Computing Machinery, Inc. https://doi.org/10.1145/3343737.3343741

DeepPlace: Learning to place applications in multi-tenant clusters. / Mitra, Subrata; Mondal, Shanka Subhra; Sheoran, Nikhil et al.
Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019. Association for Computing Machinery, Inc, 2019. p. 61-68 (Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Mitra, S, Mondal, SS, Sheoran, N, Dhake, N, Nehra, R & Simha, R 2019, DeepPlace: Learning to place applications in multi-tenant clusters. in Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019. Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019, Association for Computing Machinery, Inc, pp. 61-68, 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019, Hangzhou, China, 8/19/19. https://doi.org/10.1145/3343737.3343741

Mitra S, Mondal SS, Sheoran N, Dhake N, Nehra R, Simha R. DeepPlace: Learning to place applications in multi-tenant clusters. In Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019. Association for Computing Machinery, Inc. 2019. p. 61-68. (Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019). doi: 10.1145/3343737.3343741

@inproceedings{46b0fbfe03394021b66a1fbcb6f230f7,

title = "DeepPlace: Learning to place applications in multi-tenant clusters",

abstract = "Large multi-tenant production clusters often have to handle a variety of jobs and applications with a variety of complex resource usage characteristics. It is non-trivial and non-optimal to manually create placement rules for scheduling that would decide which applications should co-locate. In this paper, we present DeepPlace, a scheduler that learns to exploits various temporal resource usage patterns of applications using Deep Reinforcement Learning (Deep RL) to reduce resource competition across jobs running in the same machine while at the same time optimizing for overall cluster utilization.",

author = "Subrata Mitra and Mondal, \{Shanka Subhra\} and Nikhil Sheoran and Neeraj Dhake and Ravinder Nehra and Ramanuja Simha",

note = "Publisher Copyright: {\textcopyright} 2019 Copyright held by the owner/author(s). Publication rights licensedto ACM.; 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019 ; Conference date: 19-08-2019 Through 20-08-2019",

year = "2019",

month = aug,

day = "19",

doi = "10.1145/3343737.3343741",

language = "English (US)",

series = "Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019",

publisher = "Association for Computing Machinery, Inc",

pages = "61--68",

booktitle = "Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019",

}

TY - GEN

T1 - DeepPlace

T2 - 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019

AU - Mitra, Subrata

AU - Mondal, Shanka Subhra

AU - Sheoran, Nikhil

AU - Dhake, Neeraj

AU - Nehra, Ravinder

AU - Simha, Ramanuja

PY - 2019/8/19

Y1 - 2019/8/19

N2 - Large multi-tenant production clusters often have to handle a variety of jobs and applications with a variety of complex resource usage characteristics. It is non-trivial and non-optimal to manually create placement rules for scheduling that would decide which applications should co-locate. In this paper, we present DeepPlace, a scheduler that learns to exploits various temporal resource usage patterns of applications using Deep Reinforcement Learning (Deep RL) to reduce resource competition across jobs running in the same machine while at the same time optimizing for overall cluster utilization.

AB - Large multi-tenant production clusters often have to handle a variety of jobs and applications with a variety of complex resource usage characteristics. It is non-trivial and non-optimal to manually create placement rules for scheduling that would decide which applications should co-locate. In this paper, we present DeepPlace, a scheduler that learns to exploits various temporal resource usage patterns of applications using Deep Reinforcement Learning (Deep RL) to reduce resource competition across jobs running in the same machine while at the same time optimizing for overall cluster utilization.

UR - https://www.scopus.com/pages/publications/85076476114

UR - https://www.scopus.com/pages/publications/85076476114#tab=citedBy

U2 - 10.1145/3343737.3343741

DO - 10.1145/3343737.3343741

M3 - Conference contribution

AN - SCOPUS:85076476114

T3 - Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019

SP - 61

EP - 68

BT - Proceedings of the 10th ACM SIGOPS Asia-Pacific Workshop on Systems, APSys 2019

PB - Association for Computing Machinery, Inc

Y2 - 19 August 2019 through 20 August 2019

ER -

DeepPlace: Learning to place applications in multi-tenant clusters

Abstract

Publication series

Conference

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this