An experimental evaluation of datacenter workloads on low-power embedded micro servers

Yiran Zhao, Shen Li, Shaohan Hu, Hongwei Wang, Shuochao Yao, Huajie Shao, Tarek Abdelzaher

Research output: Chapter in Book/Report/Conference proceedingChapter

Abstract

This paper presents a comprehensive evaluation of an ultra- low power cluster, built upon the Intel Edison based micro servers. The improved performance and high energy effi- ciency of micro servers have driven both academia and in- dustry to explore the possibility of replacing conventional brawny servers with a larger swarm of embedded micro ser- vers. Existing attempts mostly focus on mobile-class mi- cro servers, whose capacities are similar to mobile phones. We, on the other hand, target on sensor-class micro servers, which are originally intended for uses in wearable technolo- gies, sensor networks, and Internet-of-Things. Although sensor-class micro servers have much less capacity, they are touted for minimal power consumption (< 1 Watt), which opens new possibilities of achieving higher energy efficiency in datacenter workloads. Our systematic evaluation of the Edison cluster and comparisons to conventional brawny clus- ters involve careful workload choosing and laborious param- eter tuning, which ensures maximum server utilization and thus fair comparisons. Results show that the Edison clus- ter achieves up to 3:5× improvement on work-done-per-joule for web service applications and data-intensive MapReduce jobs. In terms of scalability, the Edison cluster scales lin- early on the throughput of web service workloads, and also shows satisfactory scalability for MapReduce workloads de- spite coordination overhead.

Original languageEnglish (US)
Title of host publicationProceedings of the VLDB Endowment
PublisherAssociation for Computing Machinery
Pages696-707
Number of pages12
Edition9
DOIs
StatePublished - Jan 1 2016
Event42nd International Conference on Very Large Data Bases, VLDB 2016 - Delhi, India
Duration: Sep 5 2016Sep 9 2016

Publication series

NameProceedings of the VLDB Endowment
Number9
Volume9
ISSN (Electronic)2150-8097

Other

Other42nd International Conference on Very Large Data Bases, VLDB 2016
CountryIndia
CityDelhi
Period9/5/169/9/16

Fingerprint

Servers
Web services
Energy efficiency
Scalability
Sensors
Mobile phones
Sensor networks
Electric power utilization
Tuning
Throughput
Industry

ASJC Scopus subject areas

  • Computer Science (miscellaneous)
  • Computer Science(all)

Cite this

Zhao, Y., Li, S., Hu, S., Wang, H., Yao, S., Shao, H., & Abdelzaher, T. (2016). An experimental evaluation of datacenter workloads on low-power embedded micro servers. In Proceedings of the VLDB Endowment (9 ed., pp. 696-707). (Proceedings of the VLDB Endowment; Vol. 9, No. 9). Association for Computing Machinery. https://doi.org/10.14778/2947618.2947625

An experimental evaluation of datacenter workloads on low-power embedded micro servers. / Zhao, Yiran; Li, Shen; Hu, Shaohan; Wang, Hongwei; Yao, Shuochao; Shao, Huajie; Abdelzaher, Tarek.

Proceedings of the VLDB Endowment. 9. ed. Association for Computing Machinery, 2016. p. 696-707 (Proceedings of the VLDB Endowment; Vol. 9, No. 9).

Research output: Chapter in Book/Report/Conference proceedingChapter

Zhao, Y, Li, S, Hu, S, Wang, H, Yao, S, Shao, H & Abdelzaher, T 2016, An experimental evaluation of datacenter workloads on low-power embedded micro servers. in Proceedings of the VLDB Endowment. 9 edn, Proceedings of the VLDB Endowment, no. 9, vol. 9, Association for Computing Machinery, pp. 696-707, 42nd International Conference on Very Large Data Bases, VLDB 2016, Delhi, India, 9/5/16. https://doi.org/10.14778/2947618.2947625
Zhao Y, Li S, Hu S, Wang H, Yao S, Shao H et al. An experimental evaluation of datacenter workloads on low-power embedded micro servers. In Proceedings of the VLDB Endowment. 9 ed. Association for Computing Machinery. 2016. p. 696-707. (Proceedings of the VLDB Endowment; 9). https://doi.org/10.14778/2947618.2947625
Zhao, Yiran ; Li, Shen ; Hu, Shaohan ; Wang, Hongwei ; Yao, Shuochao ; Shao, Huajie ; Abdelzaher, Tarek. / An experimental evaluation of datacenter workloads on low-power embedded micro servers. Proceedings of the VLDB Endowment. 9. ed. Association for Computing Machinery, 2016. pp. 696-707 (Proceedings of the VLDB Endowment; 9).
@inbook{ae0157a77f40473dacdc2b3fa64bdd00,
title = "An experimental evaluation of datacenter workloads on low-power embedded micro servers",
abstract = "This paper presents a comprehensive evaluation of an ultra- low power cluster, built upon the Intel Edison based micro servers. The improved performance and high energy effi- ciency of micro servers have driven both academia and in- dustry to explore the possibility of replacing conventional brawny servers with a larger swarm of embedded micro ser- vers. Existing attempts mostly focus on mobile-class mi- cro servers, whose capacities are similar to mobile phones. We, on the other hand, target on sensor-class micro servers, which are originally intended for uses in wearable technolo- gies, sensor networks, and Internet-of-Things. Although sensor-class micro servers have much less capacity, they are touted for minimal power consumption (< 1 Watt), which opens new possibilities of achieving higher energy efficiency in datacenter workloads. Our systematic evaluation of the Edison cluster and comparisons to conventional brawny clus- ters involve careful workload choosing and laborious param- eter tuning, which ensures maximum server utilization and thus fair comparisons. Results show that the Edison clus- ter achieves up to 3:5× improvement on work-done-per-joule for web service applications and data-intensive MapReduce jobs. In terms of scalability, the Edison cluster scales lin- early on the throughput of web service workloads, and also shows satisfactory scalability for MapReduce workloads de- spite coordination overhead.",
author = "Yiran Zhao and Shen Li and Shaohan Hu and Hongwei Wang and Shuochao Yao and Huajie Shao and Tarek Abdelzaher",
year = "2016",
month = "1",
day = "1",
doi = "10.14778/2947618.2947625",
language = "English (US)",
series = "Proceedings of the VLDB Endowment",
publisher = "Association for Computing Machinery",
number = "9",
pages = "696--707",
booktitle = "Proceedings of the VLDB Endowment",
edition = "9",

}

TY - CHAP

T1 - An experimental evaluation of datacenter workloads on low-power embedded micro servers

AU - Zhao, Yiran

AU - Li, Shen

AU - Hu, Shaohan

AU - Wang, Hongwei

AU - Yao, Shuochao

AU - Shao, Huajie

AU - Abdelzaher, Tarek

PY - 2016/1/1

Y1 - 2016/1/1

N2 - This paper presents a comprehensive evaluation of an ultra- low power cluster, built upon the Intel Edison based micro servers. The improved performance and high energy effi- ciency of micro servers have driven both academia and in- dustry to explore the possibility of replacing conventional brawny servers with a larger swarm of embedded micro ser- vers. Existing attempts mostly focus on mobile-class mi- cro servers, whose capacities are similar to mobile phones. We, on the other hand, target on sensor-class micro servers, which are originally intended for uses in wearable technolo- gies, sensor networks, and Internet-of-Things. Although sensor-class micro servers have much less capacity, they are touted for minimal power consumption (< 1 Watt), which opens new possibilities of achieving higher energy efficiency in datacenter workloads. Our systematic evaluation of the Edison cluster and comparisons to conventional brawny clus- ters involve careful workload choosing and laborious param- eter tuning, which ensures maximum server utilization and thus fair comparisons. Results show that the Edison clus- ter achieves up to 3:5× improvement on work-done-per-joule for web service applications and data-intensive MapReduce jobs. In terms of scalability, the Edison cluster scales lin- early on the throughput of web service workloads, and also shows satisfactory scalability for MapReduce workloads de- spite coordination overhead.

AB - This paper presents a comprehensive evaluation of an ultra- low power cluster, built upon the Intel Edison based micro servers. The improved performance and high energy effi- ciency of micro servers have driven both academia and in- dustry to explore the possibility of replacing conventional brawny servers with a larger swarm of embedded micro ser- vers. Existing attempts mostly focus on mobile-class mi- cro servers, whose capacities are similar to mobile phones. We, on the other hand, target on sensor-class micro servers, which are originally intended for uses in wearable technolo- gies, sensor networks, and Internet-of-Things. Although sensor-class micro servers have much less capacity, they are touted for minimal power consumption (< 1 Watt), which opens new possibilities of achieving higher energy efficiency in datacenter workloads. Our systematic evaluation of the Edison cluster and comparisons to conventional brawny clus- ters involve careful workload choosing and laborious param- eter tuning, which ensures maximum server utilization and thus fair comparisons. Results show that the Edison clus- ter achieves up to 3:5× improvement on work-done-per-joule for web service applications and data-intensive MapReduce jobs. In terms of scalability, the Edison cluster scales lin- early on the throughput of web service workloads, and also shows satisfactory scalability for MapReduce workloads de- spite coordination overhead.

UR - http://www.scopus.com/inward/record.url?scp=84975853353&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84975853353&partnerID=8YFLogxK

U2 - 10.14778/2947618.2947625

DO - 10.14778/2947618.2947625

M3 - Chapter

AN - SCOPUS:84975853353

T3 - Proceedings of the VLDB Endowment

SP - 696

EP - 707

BT - Proceedings of the VLDB Endowment

PB - Association for Computing Machinery

ER -