【摘要】：數(shù)據(jù)去重能消除備份中的冗余數(shù)據(jù),節(jié)省存儲(chǔ)資源和網(wǎng)絡(luò)帶寬,因而成為當(dāng)前數(shù)據(jù)存儲(chǔ)領(lǐng)域的研究熱點(diǎn)。針對(duì)常用的塊級(jí)數(shù)據(jù)去重技術(shù)指紋查詢開銷高、系統(tǒng)吞吐率低等問題,提出一種批處理塊級(jí)數(shù)據(jù)去重方法,通過內(nèi)存緩沖區(qū)對(duì)指紋進(jìn)行排序,實(shí)現(xiàn)磁盤索引的順序查詢。同時(shí)文件以一種雙指針有向無環(huán)圖的結(jié)構(gòu)存儲(chǔ)在系統(tǒng)中,以消除文件讀時(shí)引起的隨機(jī)磁盤I/O開銷。實(shí)驗(yàn)結(jié)果表明,該方法有效克服了指紋查詢的磁盤I/O瓶頸,提高了數(shù)據(jù)去重時(shí)的系統(tǒng)讀寫性能。
[Abstract]:Data deduplication can eliminate redundant data in backup, save storage resources and network bandwidth, so it has become a research hotspot in the field of data storage. Aiming at the problems such as high overhead of fingerprint query and low throughput, a batch processing block level data removal method is proposed to sort fingerprint through memory buffer to realize sequential query of disk index. At the same time, the files are stored in the system as a double-pointer directed acyclic graph to eliminate the I / O overhead caused by random disks when the files are read. The experimental results show that the proposed method overcomes the I / O bottleneck of fingerprint query and improves the performance of reading and writing of the system.
【作者單位】： 黃淮學(xué)院國際學(xué)院;
【基金】：河南省科技項(xiàng)目(142300410288,132400411178) 河南省教育廳科學(xué)技術(shù)研究重點(diǎn)項(xiàng)目(112102210446)
【分類號(hào)】：TP333;TP309.3
，

本文編號(hào)：2242842