【摘要】：云存儲(chǔ)是一種基于互聯(lián)網(wǎng)的全新存儲(chǔ)模式，為人們提供高性價(jià)比和便捷的存儲(chǔ)服務(wù)，但其安全性是被廣泛關(guān)注和研究的熱點(diǎn)。為了保證數(shù)據(jù)的可靠性和完整性，云存儲(chǔ)中主要采取兩種措施來保證用戶數(shù)據(jù)的安全，一方面將用戶的數(shù)據(jù)冗余保存，防止因軟硬件原因致使的數(shù)據(jù)丟失；另一方面是向用戶提供數(shù)據(jù)完整性驗(yàn)證的服務(wù)，并且在發(fā)現(xiàn)存儲(chǔ)系統(tǒng)中有存儲(chǔ)節(jié)點(diǎn)出錯(cuò)或者失效時(shí)能高效的恢復(fù)出錯(cuò)數(shù)據(jù)。 目前，冗余存儲(chǔ)主要包括多副本和糾刪碼兩種存儲(chǔ)策略。多副本冗余存儲(chǔ)策略設(shè)計(jì)簡(jiǎn)單、支持高并發(fā)訪問，但需要付出成倍的空間消耗代價(jià)；糾刪碼冗余存儲(chǔ)策略容錯(cuò)能力強(qiáng)、空間利用率高，但編碼和譯碼帶來的計(jì)算開銷和訪問延遲降低了用戶的體驗(yàn)值。本文首先分析單一冗余存儲(chǔ)策略不足，提出一種基于糾刪碼的動(dòng)態(tài)副本冗余存儲(chǔ)方案(Dynamic Replication Based Erasure Codes，DRBEC)，在糾刪碼策略的基礎(chǔ)上使用副本策略�？紤]到文件修復(fù)帶寬的開銷，，該方案采用再生碼作為糾刪碼的編碼方案，將文件進(jìn)行再生碼編碼存儲(chǔ)，并根據(jù)曲線擬合預(yù)測(cè)的文件訪問熱度，動(dòng)態(tài)生成和調(diào)整文件的副本數(shù)量，發(fā)揮多副本I/O吞吐性能高的優(yōu)勢(shì)。其次，對(duì)于處于低動(dòng)態(tài)狀態(tài)的歸檔數(shù)據(jù)，本文將再生碼與MD5結(jié)合，利用MD5的唯一性給出并實(shí)現(xiàn)再生碼數(shù)據(jù)完整性驗(yàn)證和數(shù)據(jù)恢復(fù)方案，對(duì)每個(gè)再生碼分片計(jì)算其MD5值并加密隨機(jī)保存在該文件的各個(gè)數(shù)據(jù)節(jié)點(diǎn)上，使用戶無需下載原文便可實(shí)現(xiàn)對(duì)遠(yuǎn)端數(shù)據(jù)的完整性驗(yàn)證。 最后，基于Xen虛擬機(jī)搭建集群存儲(chǔ)實(shí)驗(yàn)原型系統(tǒng)，對(duì)DRBEC方案的存儲(chǔ)空間消耗、訪問性能以及再生碼編碼下基于MD5實(shí)現(xiàn)的數(shù)據(jù)完整性驗(yàn)證方案的可行性和可靠性等進(jìn)行實(shí)驗(yàn)分析。結(jié)果表明，混合冗余方案空間利用率高、平均訪問延遲低，并且提高用戶訪問的成功率；同時(shí)基于MD5的再生碼數(shù)據(jù)完整性性驗(yàn)證方案有效可靠，降低了存儲(chǔ)開銷和通信開銷，并且準(zhǔn)確定位失效節(jié)點(diǎn)位置，在低帶寬的情況下有效恢復(fù)出錯(cuò)數(shù)據(jù)，保障了數(shù)據(jù)的完整性和有效性。
[Abstract]:Cloud storage is a new storage mode based on Internet, which provides people with high performance-price ratio and convenient storage service, but its security is the focus of extensive attention and research. In order to ensure the reliability and integrity of data, two main measures are taken to ensure the security of user data in cloud storage. On the one hand, the redundant data of users is saved to prevent the loss of data caused by software and hardware. On the other hand, it provides users with the service of data integrity verification, and can recover the error data efficiently when it is found that there are errors or failures of storage nodes in the storage system. At present, redundant storage mainly includes two storage strategies: multi-copy and rectified code. The design of multi-replica redundant storage policy is simple and supports high concurrent access, but it needs to pay twice the cost of space consumption. Erasure code redundant storage policy has strong fault-tolerant ability and high spatial utilization, but the computational overhead and access delay caused by coding and decoding reduce the experience value of users. In this paper, the shortcomings of single redundant storage strategy are analyzed, and a dynamic replica redundant storage scheme based on erasure code (Dynamic Replication Based Erasure Codes,DRBEC) is proposed, and the replica strategy is used on the basis of erasure code strategy. Considering the cost of file repair bandwidth, the scheme uses regenerated code as the coding scheme of erasure code, stores the file code coding, and dynamically generates and adjusts the number of copies of the file according to the file access heat predicted by curve fitting. Give full play to the advantages of multi-copy I 鈮

本文編號(hào)：2478526