{"title":"OPS: an optimized partial stripe write scheme to improve performance of XOR-based disk arrays tolerating triple disk failures","authors":"Xunsong Huang, Chentao Wu, Jie Li","doi":"10.1145/3318265.3318274","DOIUrl":null,"url":null,"abstract":"In cloud storage and big data processing systems, RAID especially disk arrays tolerating triple disk failures (3DFTs) is a popular choice to provide high reliability with low monetary cost. For 3DFTs, a key obstacle is the low partial stripe write performance, which is caused by large amount of parity modifications based on complex erasure coding layouts. In order to solve this problem, in this paper, we propose an optimized partial stripe write (OPS) method, which reorganizes the distribution of write data blocks to share partial parities among data blocks, thereby improving overall I/O performance. The OPS method can effectively reduce the number of modified parities. To illustrate the effectiveness of our OPS method, we used Disksim to evaluate several different partial stripe write methods through simulation. The results show that OPS can reduce the average response time by up to 37.21% and decreases the number of write operations by up to 26.22% compared to the traditional partial strip writing method..","PeriodicalId":241692,"journal":{"name":"Proceedings of the 3rd International Conference on High Performance Compilation, Computing and Communications","volume":"33 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2019-03-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings of the 3rd International Conference on High Performance Compilation, Computing and Communications","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/3318265.3318274","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
In cloud storage and big data processing systems, RAID especially disk arrays tolerating triple disk failures (3DFTs) is a popular choice to provide high reliability with low monetary cost. For 3DFTs, a key obstacle is the low partial stripe write performance, which is caused by large amount of parity modifications based on complex erasure coding layouts. In order to solve this problem, in this paper, we propose an optimized partial stripe write (OPS) method, which reorganizes the distribution of write data blocks to share partial parities among data blocks, thereby improving overall I/O performance. The OPS method can effectively reduce the number of modified parities. To illustrate the effectiveness of our OPS method, we used Disksim to evaluate several different partial stripe write methods through simulation. The results show that OPS can reduce the average response time by up to 37.21% and decreases the number of write operations by up to 26.22% compared to the traditional partial strip writing method..