Infrastructure-Aware TensorFlow for Heterogeneous Datacenters

Moiz Arif; M. Mustafa Rafique; Seung Hwan Lim; Zaki Malik

doi:10.1109/MASCOTS50786.2020.9285969

Infrastructure-Aware TensorFlow for Heterogeneous Datacenters

Moiz Arif
, M. Mustafa Rafique
, Seung Hwan Lim
, Zaki Malik

Discrete Algorithms

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

5 Scopus citations

Abstract

Heterogeneous datacenters, with a variety of compute, memory, and network resources, are becoming increasingly popular to address the resource requirements of time-sensitive applications. One such application framework is the TensorFlow platform, which has become a platform of choice for running machine learning workloads. The state-of-the-art TensorFlow platform is oblivious to the availability and performance profiles of the underlying datacenter resources and does not incorporate resource requirements of the given workloads for distributed training. This leads to executing the training tasks on busy and resource-constrained worker nodes, which results in a significant increase in the overall training time. In this paper, we address this challenge and propose architectural improvements and new software modules in the default TensorFlow platform to make it aware of the availability and capabilities of the underlying datacenter resources. The proposed Infrastructure-Aware Tensor-Flow efficiently schedules the training tasks on the best possible resources for execution and reduces the overall training time. Our evaluation using the worker nodes with varying availability and performance profiles shows that the proposed enhancements yield up to 54 % reduced training time as compared to the default TensorFlow platform.

Original language	English
Title of host publication	Proceedings - 2020 IEEE 28th International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020
Publisher	IEEE Computer Society
ISBN (Electronic)	9781728192383
DOIs	https://doi.org/10.1109/MASCOTS50786.2020.9285969
State	Published - Nov 17 2020
Event	28th IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020 - Nice, France Duration: Nov 17 2020 → Nov 18 2020

Publication series

Name	Proceedings - IEEE Computer Society's Annual International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunications Systems, MASCOTS
Volume	2020-November
ISSN (Print)	1526-7539

Conference

Conference	28th IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020
Country/Territory	France
City	Nice
Period	11/17/20 → 11/18/20

Funding

Results presented in this paper are obtained using the Chameleon and CloudLab testbeds supported by the National Science Foundation. This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy (DOE). The U.S. Government retains and the publisher, by accepting the article for publication, acknowledges that the U.S. Government retains a non-exclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for U.S. Government purposes. The DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (https://www.energy.gov/downloads/doe-public-access-plan).

Keywords

Distributed TensorFlow
datacenter resource management
datacenter utilization
heterogeneous datacen-ters

Access to Document

10.1109/MASCOTS50786.2020.9285969

Cite this

Arif, M., Rafique, M. M., Lim, S. H., & Malik, Z. (2020). Infrastructure-Aware TensorFlow for Heterogeneous Datacenters. In Proceedings - 2020 IEEE 28th International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020 Article 9285969 (Proceedings - IEEE Computer Society's Annual International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunications Systems, MASCOTS; Vol. 2020-November). IEEE Computer Society. https://doi.org/10.1109/MASCOTS50786.2020.9285969

Arif, Moiz ; Rafique, M. Mustafa ; Lim, Seung Hwan et al. / Infrastructure-Aware TensorFlow for Heterogeneous Datacenters. Proceedings - 2020 IEEE 28th International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020. IEEE Computer Society, 2020. (Proceedings - IEEE Computer Society's Annual International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunications Systems, MASCOTS).

@inproceedings{fcab09cfbb464aadab5e975c32209f0d,

title = "Infrastructure-Aware TensorFlow for Heterogeneous Datacenters",

abstract = "Heterogeneous datacenters, with a variety of compute, memory, and network resources, are becoming increasingly popular to address the resource requirements of time-sensitive applications. One such application framework is the TensorFlow platform, which has become a platform of choice for running machine learning workloads. The state-of-the-art TensorFlow platform is oblivious to the availability and performance profiles of the underlying datacenter resources and does not incorporate resource requirements of the given workloads for distributed training. This leads to executing the training tasks on busy and resource-constrained worker nodes, which results in a significant increase in the overall training time. In this paper, we address this challenge and propose architectural improvements and new software modules in the default TensorFlow platform to make it aware of the availability and capabilities of the underlying datacenter resources. The proposed Infrastructure-Aware Tensor-Flow efficiently schedules the training tasks on the best possible resources for execution and reduces the overall training time. Our evaluation using the worker nodes with varying availability and performance profiles shows that the proposed enhancements yield up to 54 \% reduced training time as compared to the default TensorFlow platform.",

keywords = "Distributed TensorFlow, datacenter resource management, datacenter utilization, heterogeneous datacen-ters",

author = "Moiz Arif and Rafique, \{M. Mustafa\} and Lim, \{Seung Hwan\} and Zaki Malik",

note = "Publisher Copyright: {\textcopyright} 2020 IEEE.; 28th IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020 ; Conference date: 17-11-2020 Through 18-11-2020",

year = "2020",

month = nov,

day = "17",

doi = "10.1109/MASCOTS50786.2020.9285969",

language = "English",

series = "Proceedings - IEEE Computer Society's Annual International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunications Systems, MASCOTS",

publisher = "IEEE Computer Society",

booktitle = "Proceedings - 2020 IEEE 28th International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020",

}

Arif, M, Rafique, MM, Lim, SH & Malik, Z 2020, Infrastructure-Aware TensorFlow for Heterogeneous Datacenters. in Proceedings - 2020 IEEE 28th International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020., 9285969, Proceedings - IEEE Computer Society's Annual International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunications Systems, MASCOTS, vol. 2020-November, IEEE Computer Society, 28th IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020, Nice, France, 11/17/20. https://doi.org/10.1109/MASCOTS50786.2020.9285969

Infrastructure-Aware TensorFlow for Heterogeneous Datacenters. / Arif, Moiz; Rafique, M. Mustafa; Lim, Seung Hwan et al.
Proceedings - 2020 IEEE 28th International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020. IEEE Computer Society, 2020. 9285969 (Proceedings - IEEE Computer Society's Annual International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunications Systems, MASCOTS; Vol. 2020-November).

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

TY - GEN

T1 - Infrastructure-Aware TensorFlow for Heterogeneous Datacenters

AU - Arif, Moiz

AU - Rafique, M. Mustafa

AU - Lim, Seung Hwan

AU - Malik, Zaki

PY - 2020/11/17

Y1 - 2020/11/17

N2 - Heterogeneous datacenters, with a variety of compute, memory, and network resources, are becoming increasingly popular to address the resource requirements of time-sensitive applications. One such application framework is the TensorFlow platform, which has become a platform of choice for running machine learning workloads. The state-of-the-art TensorFlow platform is oblivious to the availability and performance profiles of the underlying datacenter resources and does not incorporate resource requirements of the given workloads for distributed training. This leads to executing the training tasks on busy and resource-constrained worker nodes, which results in a significant increase in the overall training time. In this paper, we address this challenge and propose architectural improvements and new software modules in the default TensorFlow platform to make it aware of the availability and capabilities of the underlying datacenter resources. The proposed Infrastructure-Aware Tensor-Flow efficiently schedules the training tasks on the best possible resources for execution and reduces the overall training time. Our evaluation using the worker nodes with varying availability and performance profiles shows that the proposed enhancements yield up to 54 % reduced training time as compared to the default TensorFlow platform.

AB - Heterogeneous datacenters, with a variety of compute, memory, and network resources, are becoming increasingly popular to address the resource requirements of time-sensitive applications. One such application framework is the TensorFlow platform, which has become a platform of choice for running machine learning workloads. The state-of-the-art TensorFlow platform is oblivious to the availability and performance profiles of the underlying datacenter resources and does not incorporate resource requirements of the given workloads for distributed training. This leads to executing the training tasks on busy and resource-constrained worker nodes, which results in a significant increase in the overall training time. In this paper, we address this challenge and propose architectural improvements and new software modules in the default TensorFlow platform to make it aware of the availability and capabilities of the underlying datacenter resources. The proposed Infrastructure-Aware Tensor-Flow efficiently schedules the training tasks on the best possible resources for execution and reduces the overall training time. Our evaluation using the worker nodes with varying availability and performance profiles shows that the proposed enhancements yield up to 54 % reduced training time as compared to the default TensorFlow platform.

KW - Distributed TensorFlow

KW - datacenter resource management

KW - datacenter utilization

KW - heterogeneous datacen-ters

UR - https://www.scopus.com/pages/publications/85098857856

U2 - 10.1109/MASCOTS50786.2020.9285969

DO - 10.1109/MASCOTS50786.2020.9285969

M3 - Conference contribution

AN - SCOPUS:85098857856

T3 - Proceedings - IEEE Computer Society's Annual International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunications Systems, MASCOTS

BT - Proceedings - 2020 IEEE 28th International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020

PB - IEEE Computer Society

T2 - 28th IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020

Y2 - 17 November 2020 through 18 November 2020

ER -

Arif M, Rafique MM, Lim SH, Malik Z. Infrastructure-Aware TensorFlow for Heterogeneous Datacenters. In Proceedings - 2020 IEEE 28th International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2020. IEEE Computer Society. 2020. 9285969. (Proceedings - IEEE Computer Society's Annual International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunications Systems, MASCOTS). doi: 10.1109/MASCOTS50786.2020.9285969

Infrastructure-Aware TensorFlow for Heterogeneous Datacenters

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this