Premium
Effectiveness of server load estimation by using requested file size for web server clusters
Author(s) -
Satake Shinsuke,
Inai Hiroshi,
Arai Tsuyoshi
Publication year - 2011
Publication title -
electronics and communications in japan
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.131
H-Index - 13
eISSN - 1942-9541
pISSN - 1942-9533
DOI - 10.1002/ecj.10274
Subject(s) - computer science , load balancing (electrical power) , round robin dns , polling , file server , server farm , web server , network load balancing services , web log analysis software , server , operating system , client–server model , load management , real time computing , the internet , static web page , engineering , grid , geometry , mathematics , domain name system , electrical engineering
Several dynamic load‐balancing methods for web server clusters have been proposed. In order to realize effective load‐balancing, they have required almost real‐time server load information, which could be obtained by observation of client server interactions or by frequent polling from the load‐balancer (dispatcher) to each server. However, observation provides indirect server load information and the polling interval produces a time lag. To overcome that problem, this paper proposes a load‐balancing method based on server load estimation. Under our method, the load‐balancer selects the estimated least‐loaded server, then delivers arriving requests to that server. The estimation is based on the fact that the file transfer delay is proportional to the file size. The fact makes it possible to estimate the load on each server from the last request delivery time and the file size. A performance comparison with the previously proposed methods indicates that our method can achieve an effective load balancing of web server clusters. © 2011 Wiley Periodicals, Inc. Electron Comm Jpn, 94(2): 59–66, 2011; Published online in Wiley Online Library ( wileyonlinelibrary.com ). DOI 10.1002/ecj.10274