【摘要】：云存儲(chǔ)作為一種新興的基于對(duì)象的數(shù)據(jù)存儲(chǔ)模式，，已經(jīng)成為云計(jì)算領(lǐng)域發(fā)展最早和最成熟的應(yīng)用之一。盡管云存儲(chǔ)技術(shù)具有巨大的市場(chǎng)潛力，但是在實(shí)際應(yīng)用的過(guò)程中，云存儲(chǔ)無(wú)法與傳統(tǒng)存儲(chǔ)設(shè)備和應(yīng)用進(jìn)行無(wú)縫銜接，存在訪問(wèn)接口不統(tǒng)一、訪問(wèn)延遲大等問(wèn)題。 云存儲(chǔ)網(wǎng)關(guān)系統(tǒng)實(shí)現(xiàn)了多個(gè)云存儲(chǔ)平臺(tái)接口的統(tǒng)一管理，通過(guò)緩存管理和熱點(diǎn)數(shù)據(jù)遷移降低客戶端訪問(wèn)延遲，通過(guò)選擇性價(jià)比高的云存儲(chǔ)平臺(tái)降低存儲(chǔ)成本，通過(guò)跨云平臺(tái)數(shù)據(jù)遷移實(shí)現(xiàn)云平臺(tái)之間的互操作。系統(tǒng)由4個(gè)模塊構(gòu)成，基本操作模塊監(jiān)聽并接收從客戶端發(fā)送過(guò)來(lái)的訪問(wèn)請(qǐng)求，通過(guò)用戶驗(yàn)證連接云存儲(chǔ)平臺(tái)，對(duì)請(qǐng)求進(jìn)行響應(yīng)。云平臺(tái)統(tǒng)一管理模塊將云存儲(chǔ)網(wǎng)關(guān)存儲(chǔ)池中的數(shù)據(jù)上傳到合適的云存儲(chǔ)平臺(tái)，構(gòu)建和管理云平臺(tái)與網(wǎng)關(guān)存儲(chǔ)節(jié)點(diǎn)間的文件映射表，并實(shí)現(xiàn)多個(gè)云存儲(chǔ)平臺(tái)的統(tǒng)一管理。熱點(diǎn)數(shù)據(jù)遷移模塊定時(shí)獲取存儲(chǔ)節(jié)點(diǎn)的狀態(tài)信息，根據(jù)節(jié)點(diǎn)狀態(tài)選取部分熱點(diǎn)數(shù)據(jù)，遷往合適的目的存儲(chǔ)節(jié)點(diǎn)緩存，并根據(jù)請(qǐng)求自主遷回部分熱點(diǎn)數(shù)據(jù)。跨云平臺(tái)數(shù)據(jù)遷移模塊從云存儲(chǔ)平臺(tái)上下載待遷移數(shù)據(jù)緩存至網(wǎng)關(guān)存儲(chǔ)節(jié)點(diǎn)，接著將數(shù)據(jù)上傳至目的云存儲(chǔ)平臺(tái)，然后同步文件映射表。 采用多版本的系統(tǒng)和軟件安裝鏡像以及流媒體文件，對(duì)系統(tǒng)的功能和性能進(jìn)行測(cè)試。測(cè)試結(jié)果表明，系統(tǒng)對(duì)多平臺(tái)的統(tǒng)一管理功能、熱點(diǎn)數(shù)據(jù)遷移功能、云平臺(tái)統(tǒng)一管理下載功能和跨云平臺(tái)數(shù)據(jù)遷移功能都能正常運(yùn)行。當(dāng)請(qǐng)求數(shù)據(jù)緩存在網(wǎng)關(guān)內(nèi)時(shí)，存儲(chǔ)節(jié)點(diǎn)響應(yīng)速度是云端響應(yīng)速度的5-30倍。依據(jù)阿里云和盛大云平臺(tái)的收費(fèi)標(biāo)準(zhǔn)，本系統(tǒng)的存儲(chǔ)成本僅是無(wú)網(wǎng)關(guān)系統(tǒng)的六分之一，存儲(chǔ)成本降低了86.1%。
[Abstract]:Cloud storage, as a new object based data storage mode, has become one of the earliest and most mature applications in cloud computing. Although cloud storage technology has great market potential, cloud storage can not connect seamlessly with traditional storage devices and applications in the process of practical application, and there are many problems such as disunity of access interface, large access delay and so on. Cloud storage gateway system realizes unified management of multiple cloud storage platform interfaces, reduces client access delay through cache management and hot data migration, and reduces storage cost by selecting cloud storage platform with high performance and price ratio. The interoperability between cloud platforms is realized through data migration across cloud platforms. The system is composed of four modules. The basic operation module listens on and receives the access request sent from the client, connects the cloud storage platform through the user verification, and responds to the request. The cloud platform unified management module uploads the data in the cloud storage gateway storage pool to the suitable cloud storage platform, constructs and manages the file mapping table between the cloud platform and the gateway storage node, and realizes the unified management of multiple cloud storage platforms. The hot data migration module acquires the state information of the storage node periodically, selects part of the hot data according to the node state, moves to the appropriate destination storage node cache, and moves back part of the hot spot data independently according to the request. The data migration module downloads the data to the gateway storage node from the cloud storage platform, then uploads the data to the destination cloud storage platform, and then synchronizes the file mapping table. The function and performance of the system are tested by multi-version system and software installation mirror and streaming media file. The test results show that the unified management function, hot data migration function, unified management and download function of cloud platform and data migration function across cloud platform can all run normally. When the request data is cached in the gateway, the response speed of the storage node is 5-30 times faster than that of the cloud. According to the charge standard of Ali cloud and Shanda cloud platform, the storage cost of this system is only 1/6 of that of no gateway system, and the storage cost is reduced by 86.1.
【學(xué)位授予單位】：華中科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2013
【分類號(hào)】：TP333

【參考文獻(xiàn)】

相關(guān)期刊論文 前1條

1 高建秀;吳振新;孫碩;;云存儲(chǔ)在數(shù)字資源長(zhǎng)期保存中的應(yīng)用探討[J];現(xiàn)代圖書情報(bào)技術(shù);2010年06期

本文編號(hào)：2132222