ScienceDirect - Performance Evaluation : Effective load balancing for cluster-based servers employing job preemption

Performance Evaluation
Volume 65, Issue 8, July 2008, Pages 606-622

Font Size:

Abstract - selected

Article

Figures/Tables

References

Purchase PDF (1531 K)

Article Toolbox

E-mail Article

Add to my Quick Links

Bookmark and share in 2collab (opens in new window)

Request permission to reuse this article

Cited By in Scopus (0)

Related Articles in ScienceDirect

: a load balancing policy for clustered web servers
Performance Evaluation

The RoboCup Challenge
Robotics and Autonomous Systems

Welcoming in the new breed
The New Scientist

Dynamic scheduling of stochastic jobs on a single machi...
European Journal of Operational Research

Editorial
World Patent Information

View More Related Articles

View Record in Scopus

doi:10.1016/j.peva.2008.01.001

Published by Elsevier B.V.

Effective load balancing for cluster-based servers employing job preemption

References and further reading may be available for this article. To view references and further reading you must purchase this article.

Victoria Ungureanu ^a^,, Benjamin Melamed ^b^,^, and Michael Katehakis ^c^,

^aDIMACS Center, Rutgers University, 96 Frelinghuysen Road, Piscataway, NJ 08854, United States

^bDepartment of MSIS, Rutgers University, 94 Rockafeller Road, Piscataway, NJ 08854, United States

^cDepartment of MSIS, Rutgers University, 180 University Ave., Newark, NJ 07102, United States

Received 1 September 2005;

revised 29 May 2007;

accepted 13 January 2008.

Available online 25 January 2008.

Abstract

A cluster-based server consists of a front-end dispatcher and multiple back-end servers. The dispatcher receives incoming jobs, and then decides how to assign them to back-end servers, which in turn serve the jobs according to some discipline. Cluster-based servers have been widely deployed, as they combine good performance with low costs.

Several assignment policies have been proposed for cluster-based servers, most of which aim to balance the load among back-end servers. There are two main strategies for load balancing: The first aims to balance the amount of workload at back-end servers, while the second aims to balance the number of jobs assigned to back-end servers. Examples of policies using these strategies are Dynamic and LC (Least Connected), respectively.

In this paper we propose a policy, called LC*, which combines the two aforementioned strategies. The paper shows experimentally that when preemption is admitted (i.e., when jobs execute concurrently on back-end servers), LC* substantially outperforms bothDynamic and LC in terms of response-time metrics. This improved performance is achieved by using only information readily available to the dispatcher, rendering LC* a practical policy to implement. Finally, we study a refinement, called ALC* (Adaptive LC*), which further improves on the response-time performance of LC* by adapting its actions to incoming traffic rates.

Keywords: Cluster-based servers; Back-end server architecture; Job preemption; Simulation

Article Outline

1. Introduction
2. The impact of preemption on the performance of assignment policies 2.1. Back-end server scheduling policies 2.2. Simulation data 2.3. Simulation experiments 2.3.1. Job processing without preemption 2.3.2. Job processing with preemption
3. The LC* policy
4. ALC*—an adaptive version of the LC* policy 4.1. The effect of dispatcher overhead on ALC* performance 4.2. The effect of cache misses on ALC* performance 4.3. The effect of dynamic requests on ALC* performance 4.4. Scalability of ALC*
5. Conclusion
References
Vitae

A preliminary version of this paper appeared in the Proceedings of the IEEE International Symposium on Network Computing and Applications (IEEE NCA04).

Corresponding author. Tel.: +1 732 445 3128; fax: +1 732 445 6329.

Performance Evaluation
Volume 65, Issue 8, July 2008, Pages 606-622

About ScienceDirect | Contact Us | Information for Advertisers | Terms & Conditions | Privacy Policy