{"title":"Decease I/O Mean Response Time Using Software Pipelining","authors":"Ke Zhou, Zhongying Niu","doi":"10.1109/IMSCCS.2006.52","DOIUrl":null,"url":null,"abstract":"Compared with the storage device in traditional direct attached systems, the I/O access pattern in a storage area network is different. In this paper we use fibre channel RAIDs and study how to decease I/O mean response time by software pipelining, so to enable the overlapping of different phases of the operation such as fibre channel transfer and disk access. Calculations and tests show that software pipelining can decrease I/O mean response time effectively. Further, we discuss the speedup of mean response time that is mainly influenced by two parameters: concurrent I/O request number and request data size","PeriodicalId":202629,"journal":{"name":"First International Multi-Symposiums on Computer and Computational Sciences (IMSCCS'06)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2006-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"First International Multi-Symposiums on Computer and Computational Sciences (IMSCCS'06)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/IMSCCS.2006.52","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 2
Abstract
Compared with the storage device in traditional direct attached systems, the I/O access pattern in a storage area network is different. In this paper we use fibre channel RAIDs and study how to decease I/O mean response time by software pipelining, so to enable the overlapping of different phases of the operation such as fibre channel transfer and disk access. Calculations and tests show that software pipelining can decrease I/O mean response time effectively. Further, we discuss the speedup of mean response time that is mainly influenced by two parameters: concurrent I/O request number and request data size