Previous [ 1] [ 2] [ 3] [ 4] [ 5] [ 6] [ 7] [ 8] [ 9] [ 10] [ 11] [ 12] [ 13] [ 14] [ 15] [ 16] [ 17] [ 18] [ 19]

¡@

Journal of Information Science and Engineering, Vol. 25 No. 3, pp. 747-762 (May 2009)

Strategies Toward Optimal Access to File Replicas in Data Grid Environments*

Kuan-Ching Li, Hsiao-Hsi Wang1, Kuo-Yang Cheng2 and Tsung-Ying Wu2
 Department of Computer Science and Information Engineering
1Department of Computer Science and Information Management
Providence University, 433 Taiwan
E-mail: {kuancli; hhwang}@pu.edu.tw
2National Center for High-Performance Computing
Hsinchu, 300 Taiwan
E-mail: {kuoyang; alex}@nchc.org.tw

Data Grid is one of key technologies to build up large-scale dataset storage system by connecting scattered storage resources dispersedly located in the Grid. One major challenge in data grids is how to provide good and timely access to huge amount of data in distributed locations, given the high latency of interconnection networks. Parallel downloading methods can improve download efficiency and performance, and for such, processes should have started from appropriate locations. In this research paper, we present the design framework of PU-DG Optimizer toolbox (also known as PU-DG Optibox) for data grid environments. The proposed toolbox is a package containing a number of high-end techniques and algorithms running as middleware on top of data grid platforms, in order to optimize file downloads, by improving its efficiency and performance. Moreover, PU-DG Optibox not only provides users and developers possibilities for setting their own priority strategies, as also different downloading modes. Furthermore, workload balancing to avoid low quality computing node to execute highly complex job is also included in this design. Experimental results of techniques packaged in the proposed toolbox demonstrate its potential and effectiveness.

Keywords: data grid, download optimizer toolbox, file replica selection, replica placement, hierarchy-based list

Full Text (¥þ¤åÀÉ) Retrieve PDF document (200905_06.pdf)

Received February 1, 2008; accepted November 28, 2008.
Communicated by Yau-Hwang Kuo, Pau-Choo Chung and Jar-Ferr Yang.
* This paper was based upon work supported in part by National Science Council of Taiwan, R.O.C. under grants No. NSC 95-2221-E-126-006-MY3, NSC 96-2221-E-126-004-MY3 and NSC 95-2745-E-126-002- URD.