Network Traffic Characteristics of Machine Learning Frameworks Under the Microscope
S.207-215 High computational demands of complex deep learning models led to workload distribution across multiple machines. Many frameworks for distributed machine learning (DML) have been developed and are employed in practice for orchestrating workload distribution. In this paper, we analyze and c...
Published in: | 2021 17th International Conference on Network and Service Management (CNSM) |
---|---|
Main Authors: | , , , |
Format: | Conference Object |
Language: | English |
Published: |
2021
|
Subjects: | |
Online Access: | https://publica.fraunhofer.de/handle/publica/413420 https://doi.org/10.23919/CNSM52442.2021.9615524 |
id |
ftfrauneprints:oai:publica.fraunhofer.de:publica/413420 |
---|---|
record_format |
openpolar |
spelling |
ftfrauneprints:oai:publica.fraunhofer.de:publica/413420 2024-04-21T08:01:00+00:00 Network Traffic Characteristics of Machine Learning Frameworks Under the Microscope Zerwas, Johannes Aykurt, Kaan Schmid, Stefan Blenk, Andreas 2021 https://publica.fraunhofer.de/handle/publica/413420 https://doi.org/10.23919/CNSM52442.2021.9615524 en eng International Conference on Network and Service Management (CNSM) 2021 17th International Conference on Network and Service Management, CNSM 2021. Proceedings doi:10.23919/CNSM52442.2021.9615524 https://publica.fraunhofer.de/handle/publica/413420 004 005 conference paper 2021 ftfrauneprints https://doi.org/10.23919/CNSM52442.2021.9615524 2024-03-27T15:09:44Z S.207-215 High computational demands of complex deep learning models led to workload distribution across multiple machines. Many frameworks for distributed machine learning (DML) have been developed and are employed in practice for orchestrating workload distribution. In this paper, we analyze and compare network behaviors of three widely used state-of-the-art DML frameworks. The study reveals that traffic can largely vary across the frameworks. While some frameworks exhibit well predictable patterns, others are less structured. We further explore whether and how it is possible to relate the network traffic to the DML jobs' attributes, and present a multiple linear regression model accordingly. Our results can inform the networking community about traffic characteristics and contribute toward the generation of realistic DML traffic for simulation studies. Conference Object DML Publikationsdatenbank der Fraunhofer-Gesellschaft 2021 17th International Conference on Network and Service Management (CNSM) 207 215 |
institution |
Open Polar |
collection |
Publikationsdatenbank der Fraunhofer-Gesellschaft |
op_collection_id |
ftfrauneprints |
language |
English |
topic |
004 005 |
spellingShingle |
004 005 Zerwas, Johannes Aykurt, Kaan Schmid, Stefan Blenk, Andreas Network Traffic Characteristics of Machine Learning Frameworks Under the Microscope |
topic_facet |
004 005 |
description |
S.207-215 High computational demands of complex deep learning models led to workload distribution across multiple machines. Many frameworks for distributed machine learning (DML) have been developed and are employed in practice for orchestrating workload distribution. In this paper, we analyze and compare network behaviors of three widely used state-of-the-art DML frameworks. The study reveals that traffic can largely vary across the frameworks. While some frameworks exhibit well predictable patterns, others are less structured. We further explore whether and how it is possible to relate the network traffic to the DML jobs' attributes, and present a multiple linear regression model accordingly. Our results can inform the networking community about traffic characteristics and contribute toward the generation of realistic DML traffic for simulation studies. |
format |
Conference Object |
author |
Zerwas, Johannes Aykurt, Kaan Schmid, Stefan Blenk, Andreas |
author_facet |
Zerwas, Johannes Aykurt, Kaan Schmid, Stefan Blenk, Andreas |
author_sort |
Zerwas, Johannes |
title |
Network Traffic Characteristics of Machine Learning Frameworks Under the Microscope |
title_short |
Network Traffic Characteristics of Machine Learning Frameworks Under the Microscope |
title_full |
Network Traffic Characteristics of Machine Learning Frameworks Under the Microscope |
title_fullStr |
Network Traffic Characteristics of Machine Learning Frameworks Under the Microscope |
title_full_unstemmed |
Network Traffic Characteristics of Machine Learning Frameworks Under the Microscope |
title_sort |
network traffic characteristics of machine learning frameworks under the microscope |
publishDate |
2021 |
url |
https://publica.fraunhofer.de/handle/publica/413420 https://doi.org/10.23919/CNSM52442.2021.9615524 |
genre |
DML |
genre_facet |
DML |
op_relation |
International Conference on Network and Service Management (CNSM) 2021 17th International Conference on Network and Service Management, CNSM 2021. Proceedings doi:10.23919/CNSM52442.2021.9615524 https://publica.fraunhofer.de/handle/publica/413420 |
op_doi |
https://doi.org/10.23919/CNSM52442.2021.9615524 |
container_title |
2021 17th International Conference on Network and Service Management (CNSM) |
container_start_page |
207 |
op_container_end_page |
215 |
_version_ |
1796941374618075136 |