本文選題：NAND　切入點：Flash　出處：《國防科學(xué)技術(shù)大學(xué)》2012年碩士論文

【摘要】：NAND Flash容量的增長和其性能及可靠性之間存在著矛盾。單顆NAND Flash的容量不斷增大的同時，其基本單元——塊和頁的容量也在隨之增大，但是文件系統(tǒng)訪問SSD時的基本單位——扇區(qū)沒有變。一個物理頁包含的邏輯扇區(qū)數(shù)越來越多，并且頁具有先擦再寫的特性，寫入的最小單位是一個頁，擦除的最小單位是一個塊，故重寫入問題越來越嚴重，需要探索更有效的架構(gòu)和管理算法來提高存儲的性能。并且，隨著容量的增大，NAND Flash制程進一步降低，致使NANDFlash存儲單元總的擦寫次數(shù)不斷降低，需要更加充分利用有限的擦寫次數(shù)來提高存儲系統(tǒng)的可靠性。此外，數(shù)據(jù)量的驟增對SSD的容量提出了新的要求，現(xiàn)有NAND Flash存儲組織架構(gòu)已經(jīng)不能適應(yīng)容量增長的要求。 針對當(dāng)前NAND Flash容量和性能之間，，以及物理塊/頁大小與扇區(qū)大小之間存在的矛盾，設(shè)計了一種非均衡的NAND Flash存儲組織架構(gòu)，包括非均衡通道和混合NAND Flash。非均衡通道將幾個物理通道合并形成超級通道，使NAND Flash的“物理存儲單元”進一步增大，從而降低地址映射表的規(guī)模，減小RAM資源的消耗。混合NAND Flash通過將MLC和SLC混合使用，充分利用SLC的性能和可靠性優(yōu)勢以及MLC的價格和容量優(yōu)勢來提高系統(tǒng)容量，同時保證系統(tǒng)性能�；谔岢龅姆蔷饧軜�(gòu)，還設(shè)計了基于塊組的地址映射算法、通道數(shù)據(jù)濾波器以及請求隊列的調(diào)度優(yōu)化算法。 在DiskSim和SSD Model的基礎(chǔ)上，設(shè)計了非均衡NAND Flash架構(gòu)下的SSD性能和可靠性仿真平臺。在指定的性能和可靠性評估條件下，映射表減小了25%到50%，RAM資源占用最小約為BAT的十分之一；隨機數(shù)據(jù)讀寫性能方面，提出的非均衡組織架構(gòu)以25%的SLC比例達到將近100%SLC系統(tǒng)的性能，順序讀寫速度介于全SLC和全MLC之間，增幅不大；可靠性能相對全MLC系統(tǒng)大幅提升。
[Abstract]:There is a contradiction between the growth of NAND Flash capacity and its performance and reliability. As the capacity of a single NAND Flash increases, so does the capacity of its basic unit, blocks and pages. But the basic unit of file system access to SSD-the sector remains the same. A physical page contains more and more logical sectors, and the page has the feature of erasing before writing, the smallest unit of writing is a page, and the smallest unit of erasure is a block. Therefore, the problem of rewriting is becoming more and more serious, so it is necessary to explore more effective architecture and management algorithms to improve storage performance. It is necessary to make full use of the limited erasing times to improve the reliability of the storage system. In addition, the sudden increase in the amount of data puts forward new requirements for the capacity of SSD, and the existing NAND Flash storage organization can no longer meet the requirements of capacity growth. Aiming at the contradiction between current NAND Flash capacity and performance, and between physical block / page size and sector size, an unbalanced NAND Flash storage organization architecture is designed. Including unbalanced channels and mixed NAND Flash. Unbalanced channels combine several physical channels to form superchannels, further increasing the "physical storage unit" of NAND Flash, thus reducing the size of address mapping tables. To reduce the consumption of RAM resources, hybrid NAND Flash improves system capacity by mixing MLC and SLC, making full use of the performance and reliability advantages of SLC and the price and capacity advantages of MLC, while ensuring system performance. The address mapping algorithm based on block group, channel data filter and scheduling optimization algorithm of request queue are also designed. On the basis of DiskSim and SSD Model, the simulation platform of SSD performance and reliability under unbalanced NAND Flash architecture is designed. Under the specified performance and reliability evaluation conditions, the mapping table reduces the minimum resource consumption of BAT by 25% to 50% of BAT. In the aspect of reading and writing performance of random data, the proposed disequilibrium organization architecture achieves the performance of nearly 100%SLC system by 25% SLC, the speed of sequential reading and writing is between full SLC and full MLC, the increase is not large, and the reliability performance is greatly improved compared with the whole MLC system.
【學(xué)位授予單位】：國防科學(xué)技術(shù)大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2012
【分類號】：TP333

【參考文獻】

相關(guān)期刊論文 前3條

1 黃濱;俞建新;;大容量閃存的層次型熱數(shù)據(jù)識別框架[J];計算機工程;2008年22期

2 鐘誠;岳麗華;韋鵬;;一種多路Flash存儲結(jié)構(gòu)的熱數(shù)據(jù)判斷方法[J];計算機工程;2009年12期

3 張玉芳;陽佶宏;熊忠陽;王志遠;;基于隊列計數(shù)的固態(tài)存儲器熱數(shù)據(jù)識別方法[J];計算機應(yīng)用研究;2011年08期

本文編號：1665844