The power of slightly more than one sample in randomized load balancing

Lei Ying, R. Srikant, Xiaohan Kang

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

35 Citations (Scopus)

Abstract

In many computing and networking applications, arriving tasks have to be routed to one of many servers, with the goal of minimizing queueing delays. When the number of processors is very large, a popular routing algorithm works as follows: select two servers at random and route an arriving task to the least loaded of the two. It is well-known that this algorithm dramatically reduces queueing delays compared to an algorithm which routes to a single randomly selected server. In recent cloud computing applications, it has been observed that even sampling two queues per arriving task can be expensive and can even increase delays due to messaging overhead. So there is an interest in reducing the number of sampled queues per arriving task. In this paper, we show that the number of sampled queues can be dramatically reduced by using the fact that tasks arrive in batches (called jobs). In particular, we sample a subset of the queues such that the size of the subset is slightly larger than the batch size (thus, on average, we only sample slightly more than one queue per task). Once a random subset of the queues is sampled, we propose a new load balancing method called batch-filling to attempt to equalize the load among the sampled servers. We show that our algorithm dramatically reduces the sample complexity compared to previously proposed algorithms.

Original languageEnglish (US)
Title of host publicationProceedings - IEEE INFOCOM
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1131-1139
Number of pages9
Volume26
ISBN (Print)9781479983810
DOIs
StatePublished - Aug 21 2015
Event34th IEEE Annual Conference on Computer Communications and Networks, IEEE INFOCOM 2015 - Hong Kong, Hong Kong
Duration: Apr 26 2015May 1 2015

Other

Other34th IEEE Annual Conference on Computer Communications and Networks, IEEE INFOCOM 2015
CountryHong Kong
CityHong Kong
Period4/26/155/1/15

Fingerprint

Resource allocation
Servers
Routing algorithms
Cloud computing
Sampling

ASJC Scopus subject areas

  • Computer Science(all)
  • Electrical and Electronic Engineering

Cite this

Ying, L., Srikant, R., & Kang, X. (2015). The power of slightly more than one sample in randomized load balancing. In Proceedings - IEEE INFOCOM (Vol. 26, pp. 1131-1139). [7218487] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/INFOCOM.2015.7218487

The power of slightly more than one sample in randomized load balancing. / Ying, Lei; Srikant, R.; Kang, Xiaohan.

Proceedings - IEEE INFOCOM. Vol. 26 Institute of Electrical and Electronics Engineers Inc., 2015. p. 1131-1139 7218487.

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

Ying, L, Srikant, R & Kang, X 2015, The power of slightly more than one sample in randomized load balancing. in Proceedings - IEEE INFOCOM. vol. 26, 7218487, Institute of Electrical and Electronics Engineers Inc., pp. 1131-1139, 34th IEEE Annual Conference on Computer Communications and Networks, IEEE INFOCOM 2015, Hong Kong, Hong Kong, 4/26/15. https://doi.org/10.1109/INFOCOM.2015.7218487
Ying L, Srikant R, Kang X. The power of slightly more than one sample in randomized load balancing. In Proceedings - IEEE INFOCOM. Vol. 26. Institute of Electrical and Electronics Engineers Inc. 2015. p. 1131-1139. 7218487 https://doi.org/10.1109/INFOCOM.2015.7218487
Ying, Lei ; Srikant, R. ; Kang, Xiaohan. / The power of slightly more than one sample in randomized load balancing. Proceedings - IEEE INFOCOM. Vol. 26 Institute of Electrical and Electronics Engineers Inc., 2015. pp. 1131-1139
@inproceedings{710843f755094bcd99e38d00deced343,
title = "The power of slightly more than one sample in randomized load balancing",
abstract = "In many computing and networking applications, arriving tasks have to be routed to one of many servers, with the goal of minimizing queueing delays. When the number of processors is very large, a popular routing algorithm works as follows: select two servers at random and route an arriving task to the least loaded of the two. It is well-known that this algorithm dramatically reduces queueing delays compared to an algorithm which routes to a single randomly selected server. In recent cloud computing applications, it has been observed that even sampling two queues per arriving task can be expensive and can even increase delays due to messaging overhead. So there is an interest in reducing the number of sampled queues per arriving task. In this paper, we show that the number of sampled queues can be dramatically reduced by using the fact that tasks arrive in batches (called jobs). In particular, we sample a subset of the queues such that the size of the subset is slightly larger than the batch size (thus, on average, we only sample slightly more than one queue per task). Once a random subset of the queues is sampled, we propose a new load balancing method called batch-filling to attempt to equalize the load among the sampled servers. We show that our algorithm dramatically reduces the sample complexity compared to previously proposed algorithms.",
author = "Lei Ying and R. Srikant and Xiaohan Kang",
year = "2015",
month = "8",
day = "21",
doi = "10.1109/INFOCOM.2015.7218487",
language = "English (US)",
isbn = "9781479983810",
volume = "26",
pages = "1131--1139",
booktitle = "Proceedings - IEEE INFOCOM",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - The power of slightly more than one sample in randomized load balancing

AU - Ying, Lei

AU - Srikant, R.

AU - Kang, Xiaohan

PY - 2015/8/21

Y1 - 2015/8/21

N2 - In many computing and networking applications, arriving tasks have to be routed to one of many servers, with the goal of minimizing queueing delays. When the number of processors is very large, a popular routing algorithm works as follows: select two servers at random and route an arriving task to the least loaded of the two. It is well-known that this algorithm dramatically reduces queueing delays compared to an algorithm which routes to a single randomly selected server. In recent cloud computing applications, it has been observed that even sampling two queues per arriving task can be expensive and can even increase delays due to messaging overhead. So there is an interest in reducing the number of sampled queues per arriving task. In this paper, we show that the number of sampled queues can be dramatically reduced by using the fact that tasks arrive in batches (called jobs). In particular, we sample a subset of the queues such that the size of the subset is slightly larger than the batch size (thus, on average, we only sample slightly more than one queue per task). Once a random subset of the queues is sampled, we propose a new load balancing method called batch-filling to attempt to equalize the load among the sampled servers. We show that our algorithm dramatically reduces the sample complexity compared to previously proposed algorithms.

AB - In many computing and networking applications, arriving tasks have to be routed to one of many servers, with the goal of minimizing queueing delays. When the number of processors is very large, a popular routing algorithm works as follows: select two servers at random and route an arriving task to the least loaded of the two. It is well-known that this algorithm dramatically reduces queueing delays compared to an algorithm which routes to a single randomly selected server. In recent cloud computing applications, it has been observed that even sampling two queues per arriving task can be expensive and can even increase delays due to messaging overhead. So there is an interest in reducing the number of sampled queues per arriving task. In this paper, we show that the number of sampled queues can be dramatically reduced by using the fact that tasks arrive in batches (called jobs). In particular, we sample a subset of the queues such that the size of the subset is slightly larger than the batch size (thus, on average, we only sample slightly more than one queue per task). Once a random subset of the queues is sampled, we propose a new load balancing method called batch-filling to attempt to equalize the load among the sampled servers. We show that our algorithm dramatically reduces the sample complexity compared to previously proposed algorithms.

UR - http://www.scopus.com/inward/record.url?scp=84954235294&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84954235294&partnerID=8YFLogxK

U2 - 10.1109/INFOCOM.2015.7218487

DO - 10.1109/INFOCOM.2015.7218487

M3 - Conference contribution

SN - 9781479983810

VL - 26

SP - 1131

EP - 1139

BT - Proceedings - IEEE INFOCOM

PB - Institute of Electrical and Electronics Engineers Inc.

ER -