Authors: Diqing Hu, Qiang Cao, ChangSheng Xie
Addresses: College of Computer Science and Technology, Huazhong University of Science and Technology, 1037 Luoyu St., Wuhan 430074, China. ' College of Computer Science and Technology, Huazhong University of Science and Technology, 1037 Luoyu St., Wuhan 430074, China. ' Division of Data Storage System, Wuhan National Laboratory for Optoelectronics, 1037 Luoyu St., Wuhan 430074, China
Abstract: The I/O routine in network storage has longer transferring time and overhead than local I/O. To make good use of the storage network bandwidth and meet demand of concurrent I/O requests, we need to find the appropriate schedule policy for I/O requests at the level of the network. The focus in the paper is on how to schedule concurrent I/O data transferring in share storage network, and to improve performance such as the response time affected by scheduling method. Our objective is to analyse the response time of concurrent requests as function of their schedule order such as the sequenced, parallel and hybrid policies. We then verify our results by both the analytical and experimental method.
Keywords: storage networks; input/output; I/O; schedule policy; concurrent requests; data transfer; scheduling.
International Journal of High Performance Computing and Networking, 2010 Vol.6 No.3/4, pp.208 - 212
Available online: 30 Dec 2010 *Full-text access for editors Access for subscribers Purchase this article Comment on this article