Zhou Lei, Zhebo Zhou, Bolin Hu, W. Shen, Luokai Hu
{"title":"PIOD: A Parallel I/O Dispatch Model Based on Lustre File System for Virtual Machine Storage","authors":"Zhou Lei, Zhebo Zhou, Bolin Hu, W. Shen, Luokai Hu","doi":"10.1109/CSC.2013.13","DOIUrl":null,"url":null,"abstract":"Virtualization is a key technique for cloud computing. One challenge in it is providing a scalable, high performance and high cost effective data storage for virtual machines. In this paper, we propose a parallel I/O dispatch model called PIOD based on Lustre parallel file system to address this issue. PIOD adopts queue mechanism to coordinate data access requests from clients, avoiding random and concurrent storage access. Doing so contributes to avoid random disk seek overhead, which in turn improves I/O performance of data access. The simulation results show that PIOD is effective and can improve I/O throughput about 26.3% on average.","PeriodicalId":338002,"journal":{"name":"2013 International Conference on Cloud and Service Computing","volume":"43 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2013-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2013 International Conference on Cloud and Service Computing","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/CSC.2013.13","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
Virtualization is a key technique for cloud computing. One challenge in it is providing a scalable, high performance and high cost effective data storage for virtual machines. In this paper, we propose a parallel I/O dispatch model called PIOD based on Lustre parallel file system to address this issue. PIOD adopts queue mechanism to coordinate data access requests from clients, avoiding random and concurrent storage access. Doing so contributes to avoid random disk seek overhead, which in turn improves I/O performance of data access. The simulation results show that PIOD is effective and can improve I/O throughput about 26.3% on average.