|
Title:
|
BURSTINESS DETECTION LOCATION IN CLUSTER- BASED NETWORK SERVERS BASED ON THE APRAALGORITHM: A CASE STUDY |
|
Author(s):
|
Katja Gilly , Javier Huertas Amores , Carlos Juiz , Ramon Puigjaner |
|
ISBN:
|
972-8924-19-4 |
|
Editors:
|
Pedro Isaías, Miguel Baptista Nunes and Inmaculada J. Martínez |
|
Year:
|
2005 |
|
Edition:
|
V II, 2 |
|
Keywords:
|
Web applications, burstiness, load-balancing, Web cluster. |
|
Type:
|
Short Paper |
|
First Page:
|
177 |
|
Last Page:
|
181 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
This paper presents a study of the location of burstiness detection implemented on a heuristic adaptive predictive load balancing algorithm. Our main objective is to provide an estimation of the arrival rate trend in order to prevent the system from a possible collapse when intense overload is expected. Our load balancing algorithm includes service differentiation and admission control while paying low overhead. We have implemented a prototype of the model on a Linux cluster and captured the burstiness level in several system locations. Results show that system performance depends on where the burstiness is detected considering the overload of the system and that burstiness control in back-end nodes provides useful feedback on the decisions taken by the load balancing algorithm. |
|
|
|
|
|
|
