HyLine

A Simple and Practical Flow Scheduling for Commodity Datacenters

Soheil Abbasloo, Yang Xu, H. Jonathan Chao

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Today's datacenter networks (DCNs) have been built upon multipath topologies where each path contains multiple links. However, flow scheduling schemes proposed to minimize flow completion times (FCT) in DCNs are based on algorithms which are optimum or close-to-optimum only over single link. Moreover, most of these scheduling schemes seek either fully centralized approaches having overhead of communicating to a central entity or fully distributed approaches requiring changes in the fabric.Motivated by these shortcomings, we present HyLine a simple scheduling design for commodity DCNs which is equipped with a joint load-balancing and flow scheduling (path-aware) design exploiting the multipath nature of DCNs. HyLine takes a hybrid approach and uses the global-awareness of centralized and agility of distributed techniques without requiring any changes in the fabric. To that end, it determines a threshold margin identifying flows for which using centralized approach is beneficial.We have shown through extensive ns2 simulations that despite HyLine's simplicity, it significantly outperforms existing schemes and achieves lower average and 99th percentile FCTs. For instance, compared to Qjump-state-of-the-art practical scheme- and pFabric-one of the best performing flow scheduling schemes- HyLine reduces average FCT up to 68% and 31%, respectively, under a production datacenter workload.

Original languageEnglish (US)
Title of host publication2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages334-342
Number of pages9
ISBN (Electronic)9783903176089
DOIs
StatePublished - Apr 22 2019
Event17th IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Zurich, Switzerland
Duration: May 14 2018May 16 2018

Publication series

Name2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings

Conference

Conference17th IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018
CountrySwitzerland
CityZurich
Period5/14/185/16/18

Fingerprint

Scheduling
Resource allocation
Topology

ASJC Scopus subject areas

  • Computer Networks and Communications

Cite this

Abbasloo, S., Xu, Y., & Chao, H. J. (2019). HyLine: A Simple and Practical Flow Scheduling for Commodity Datacenters. In 2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings (pp. 334-342). [8696675] (2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.23919/IFIPNetworking.2018.8696675

HyLine : A Simple and Practical Flow Scheduling for Commodity Datacenters. / Abbasloo, Soheil; Xu, Yang; Chao, H. Jonathan.

2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2019. p. 334-342 8696675 (2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abbasloo, S, Xu, Y & Chao, HJ 2019, HyLine: A Simple and Practical Flow Scheduling for Commodity Datacenters. in 2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings., 8696675, 2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings, Institute of Electrical and Electronics Engineers Inc., pp. 334-342, 17th IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018, Zurich, Switzerland, 5/14/18. https://doi.org/10.23919/IFIPNetworking.2018.8696675
Abbasloo S, Xu Y, Chao HJ. HyLine: A Simple and Practical Flow Scheduling for Commodity Datacenters. In 2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2019. p. 334-342. 8696675. (2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings). https://doi.org/10.23919/IFIPNetworking.2018.8696675
Abbasloo, Soheil ; Xu, Yang ; Chao, H. Jonathan. / HyLine : A Simple and Practical Flow Scheduling for Commodity Datacenters. 2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 334-342 (2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings).
@inproceedings{8197477c5d324a9db45c887639c43fec,
title = "HyLine: A Simple and Practical Flow Scheduling for Commodity Datacenters",
abstract = "Today's datacenter networks (DCNs) have been built upon multipath topologies where each path contains multiple links. However, flow scheduling schemes proposed to minimize flow completion times (FCT) in DCNs are based on algorithms which are optimum or close-to-optimum only over single link. Moreover, most of these scheduling schemes seek either fully centralized approaches having overhead of communicating to a central entity or fully distributed approaches requiring changes in the fabric.Motivated by these shortcomings, we present HyLine a simple scheduling design for commodity DCNs which is equipped with a joint load-balancing and flow scheduling (path-aware) design exploiting the multipath nature of DCNs. HyLine takes a hybrid approach and uses the global-awareness of centralized and agility of distributed techniques without requiring any changes in the fabric. To that end, it determines a threshold margin identifying flows for which using centralized approach is beneficial.We have shown through extensive ns2 simulations that despite HyLine's simplicity, it significantly outperforms existing schemes and achieves lower average and 99th percentile FCTs. For instance, compared to Qjump-state-of-the-art practical scheme- and pFabric-one of the best performing flow scheduling schemes- HyLine reduces average FCT up to 68{\%} and 31{\%}, respectively, under a production datacenter workload.",
author = "Soheil Abbasloo and Yang Xu and Chao, {H. Jonathan}",
year = "2019",
month = "4",
day = "22",
doi = "10.23919/IFIPNetworking.2018.8696675",
language = "English (US)",
series = "2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "334--342",
booktitle = "2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings",

}

TY - GEN

T1 - HyLine

T2 - A Simple and Practical Flow Scheduling for Commodity Datacenters

AU - Abbasloo, Soheil

AU - Xu, Yang

AU - Chao, H. Jonathan

PY - 2019/4/22

Y1 - 2019/4/22

N2 - Today's datacenter networks (DCNs) have been built upon multipath topologies where each path contains multiple links. However, flow scheduling schemes proposed to minimize flow completion times (FCT) in DCNs are based on algorithms which are optimum or close-to-optimum only over single link. Moreover, most of these scheduling schemes seek either fully centralized approaches having overhead of communicating to a central entity or fully distributed approaches requiring changes in the fabric.Motivated by these shortcomings, we present HyLine a simple scheduling design for commodity DCNs which is equipped with a joint load-balancing and flow scheduling (path-aware) design exploiting the multipath nature of DCNs. HyLine takes a hybrid approach and uses the global-awareness of centralized and agility of distributed techniques without requiring any changes in the fabric. To that end, it determines a threshold margin identifying flows for which using centralized approach is beneficial.We have shown through extensive ns2 simulations that despite HyLine's simplicity, it significantly outperforms existing schemes and achieves lower average and 99th percentile FCTs. For instance, compared to Qjump-state-of-the-art practical scheme- and pFabric-one of the best performing flow scheduling schemes- HyLine reduces average FCT up to 68% and 31%, respectively, under a production datacenter workload.

AB - Today's datacenter networks (DCNs) have been built upon multipath topologies where each path contains multiple links. However, flow scheduling schemes proposed to minimize flow completion times (FCT) in DCNs are based on algorithms which are optimum or close-to-optimum only over single link. Moreover, most of these scheduling schemes seek either fully centralized approaches having overhead of communicating to a central entity or fully distributed approaches requiring changes in the fabric.Motivated by these shortcomings, we present HyLine a simple scheduling design for commodity DCNs which is equipped with a joint load-balancing and flow scheduling (path-aware) design exploiting the multipath nature of DCNs. HyLine takes a hybrid approach and uses the global-awareness of centralized and agility of distributed techniques without requiring any changes in the fabric. To that end, it determines a threshold margin identifying flows for which using centralized approach is beneficial.We have shown through extensive ns2 simulations that despite HyLine's simplicity, it significantly outperforms existing schemes and achieves lower average and 99th percentile FCTs. For instance, compared to Qjump-state-of-the-art practical scheme- and pFabric-one of the best performing flow scheduling schemes- HyLine reduces average FCT up to 68% and 31%, respectively, under a production datacenter workload.

UR - http://www.scopus.com/inward/record.url?scp=85063300015&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85063300015&partnerID=8YFLogxK

U2 - 10.23919/IFIPNetworking.2018.8696675

DO - 10.23919/IFIPNetworking.2018.8696675

M3 - Conference contribution

T3 - 2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings

SP - 334

EP - 342

BT - 2018 IFIP Networking Conference IFIP Networking and Workshops, IFIP Networking 2018 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

ER -