存儲(chǔ)堆棧數(shù)據(jù)損壞分析

1、存儲(chǔ)堆棧中的數(shù)據(jù)損壞問題分析Lakshmi N. Bairavasundarambairavasundaram Lakshmi N., Garth R. Goodson古德森，加思R., Bianca Schroeder，比安卡施羅德Andrea C. Arpaci-Dusseau安德列C. arpaci杜索, Remzi H. Arpaci-Dusseau該arpaci杜索，H.University of Wisconsin-Madison威斯康星大學(xué)-麥迪遜Network Appliance, Inc.網(wǎng)絡(luò)設(shè)備公司University of Toronto多倫多大學(xué)laksh, dusse

2、au, , garth.goodson, 拉克，杜索，該 ，garth.goodson，Abstract摘要An important threat to reliable storage of data is silent對(duì)數(shù)據(jù)可靠存儲(chǔ)的一個(gè)重要威脅是無聲的data corruption. In order to develop suitable protection數(shù)據(jù)腐敗。為了開發(fā)合適的保護(hù)mechanisms against data corruption

3、, it is essential to understand its characteristics. In this paper, we present the對(duì)數(shù)據(jù)腐敗的機(jī)制，它是必不可少的，以了解其特點(diǎn)。在本文中，我們提出了first large-scale study of data corruption. We analyze corruption instances recorded in production storage systems第一次大規(guī)模數(shù)據(jù)腐敗研究。我們分析記錄在生產(chǎn)存儲(chǔ)系統(tǒng)的腐敗現(xiàn)象containing a total of 1.53 million disk

4、 drives, over a period of 41 months. We study three classes of corruption:包含1530000個(gè)磁盤驅(qū)動(dòng)器，超過41個(gè)月的時(shí)間。我們研究了三類腐?。篶hecksum mismatches, identity discrepancies, and parity inconsistencies. We focus on checksum mismatches校驗(yàn)和不匹配，身份的差異，和奇偶校驗(yàn)不一致。我們專注于校驗(yàn)和不匹配since they occur the most.因?yàn)樗麄冏睢e find more than 400

5、,000 instances of checksum我們發(fā)現(xiàn)校驗(yàn)和400000多個(gè)實(shí)例mismatches over the 41-month period. We find many41個(gè)月內(nèi)不匹配。我們發(fā)現(xiàn)很多interesting trends among these instances including: (i)有趣的趨勢(shì)，在這些情況下，包括：（我）nearline disks (and their adapters) develop checksum近線盤（和適配器）開發(fā)的校驗(yàn)mismatches an order of magnitude more often than ente

6、rprise class disk drives, (ii) checksum mismatches within錯(cuò)位的幅度往往比企業(yè)級(jí)磁盤驅(qū)動(dòng)器的順序，（ii）在校驗(yàn)和不匹配the same disk are not independent events and they show同一個(gè)磁盤不是獨(dú)立的事件，它們顯示high spatial and temporal locality, and (iii) checksum高的時(shí)間和空間局部性，及（iii）校驗(yàn)mismatches across different disks in the same storage在同一存儲(chǔ)的不同磁盤上的不匹配

7、system are not independent. We use our observations to系統(tǒng)不是獨(dú)立的。我們用我們的意見derive lessons for corruption-proof system design.從中吸取教訓(xùn)，以防腐敗體系設(shè)計(jì)。1 Introduction1引言O(shè)ne of the biggest challenges in designing storage systems is providing the reliability and availability that users在設(shè)計(jì)存儲(chǔ)系統(tǒng)的最大挑戰(zhàn)之一是提供的可靠性和可用性，用戶expe

8、ct. Once their data is stored, users expect it to be persistent forever, and perpetually available. Unfortunately,期待。一旦他們的數(shù)據(jù)存儲(chǔ)，用戶期望它會(huì)持續(xù)永遠(yuǎn)，永遠(yuǎn)有效。不幸的是，in practice there are a number of problems that, if not在實(shí)踐中有許多問題，如果不dealt with, can cause data loss in storage systems.處理，可引起存儲(chǔ)系統(tǒng)中的數(shù)據(jù)丟失。One primary caus

9、e of data loss is disk drive unreliability 16. It is well-known that hard drives are mechanical, moving devices that can suffer from mechanical problems leading to drive failure and data loss. For數(shù)據(jù)丟失的一個(gè)主要原因是磁盤驅(qū)動(dòng)器的可靠性 16 。眾所周知，硬盤是機(jī)械的，移動(dòng)的設(shè)備，可以承受機(jī)械故障導(dǎo)致的故障和數(shù)據(jù)丟失。對(duì)于example, media imperfections, and loose

10、 particles causing scratches, contribute to media errors, referred to as例如，媒體的不完善，以及松散的顆粒造成的劃傷，有助于媒體的錯(cuò)誤，簡稱為latent sector errors, within disk drives 18. Latent sector潛在的部門錯(cuò)誤，在磁盤驅(qū)動(dòng)器 18 。潛在部門errors are detected by a drives internal error-correcting錯(cuò)誤被檢測到驅(qū)動(dòng)器的內(nèi)部錯(cuò)誤校正codes (ECC) and are reported to the sto

11、rage system.碼（ECC）和報(bào)告存儲(chǔ)系統(tǒng)。Less well-known, however, is that current hard drives然而，眾所周知，目前的硬盤驅(qū)動(dòng)器and controllers consist of hundreds-of-thousands of lines和控制器由數(shù)百條線組成of low-level firmware code. This firmware code, along低級(jí)別固件代碼。這個(gè)固件代碼，一起with higher-level system software, has the potential for使用更高級(jí)別的系統(tǒng)軟件

12、，具有潛在的harboring bugs that can cause a more insidious type of窩藏錯(cuò)誤，可以導(dǎo)致更陰險(xiǎn)的類型disk error silent data corruption, where the data is磁盤錯(cuò)誤：數(shù)據(jù)是錯(cuò)誤的，數(shù)據(jù)是錯(cuò)誤的silently corrupted with no indication from the drive that無聲的損壞，沒有任何跡象表明，從驅(qū)動(dòng)器an error has occurred.發(fā)生錯(cuò)誤。Silent data corruptions could lead to data loss more

13、 often than latent sector errors, since, unlike latent sector errors, they cannot be detected or repaired by the disk drive靜默數(shù)據(jù)損壞可能會(huì)導(dǎo)致數(shù)據(jù)丟失的往往比潛在扇區(qū)錯(cuò)誤，因?yàn)?，不像潛在扇區(qū)錯(cuò)誤，他們無法檢測或修復(fù)的磁盤驅(qū)動(dòng)器itself. Detecting and recovering from data corruption requires protection techniques beyond those provided by本身。檢測和恢復(fù)數(shù)據(jù)損壞需要保

14、護(hù)技術(shù)，超越了那些提供the disk drive. In fact, basic protection schemes such as磁盤驅(qū)動(dòng)器。事實(shí)上，基本的保護(hù)計(jì)劃，如RAID 13 may also be unable to detect these problems.襲擊 13 可能也無法檢測到這些問題。The most common technique used in storage systems存儲(chǔ)系統(tǒng)中最常用的技術(shù)to detect data corruption is for the storage system to add檢測數(shù)據(jù)腐敗，是為存儲(chǔ)系統(tǒng)添加its own h

15、igher-level checksum for each disk block, which自己的上級(jí)校驗(yàn)每個(gè)磁盤塊，這is validated on each disk block read. There is a long history of enterprise-class storage systems, including ours,在每個(gè)磁盤塊上進(jìn)行驗(yàn)證。企業(yè)級(jí)存儲(chǔ)系統(tǒng)有很長的歷史，包括我們的，in using checksums in a variety of manners to detect data在以各種方式使用校驗(yàn)和檢測數(shù)據(jù)corruption 3, 6, 8, 2

16、2. However, as we discuss later,腐敗 3，6，8，22 。然而，我們稍后再討論，checksums do not protect against all forms of corruption.校驗(yàn)和不保護(hù)反對(duì)一切形式的腐敗。Therefore, in addition to checksums, our storage system因此，除了校驗(yàn)和，我們的存儲(chǔ)系統(tǒng)also uses file system-level disk block identity information to detect previously undetectable corrup

17、tions.使用文件系統(tǒng)級(jí)的磁盤塊的身份信息來檢測從未發(fā)現(xiàn)的腐敗。In order to further improve on techniques to handle為了進(jìn)一步提高處理技術(shù)corruption, we need to develop a thorough understanding腐敗，我們需要深入了解of data corruption characteristics. While recent studies數(shù)據(jù)腐敗特征。而最近的研究provide information on whole disk failures 11, 14, 16提供整個(gè)磁盤故障的信息 11，14

18、，16 and latent sector errors 2 that can aid system designers和潛在部門的錯(cuò)誤 2 ，可以幫助系統(tǒng)設(shè)計(jì)師in handling these error conditions, very little is known在處理這些錯(cuò)誤的情況下，很少是已知的about data corruption, its prevalence and its characteristics. This paper presents a large-scale study of silent關(guān)于數(shù)據(jù)腐敗，其患病率及其特點(diǎn)。本文提出了一種大規(guī)模的研究，沉默d

19、ata corruption based on field data from 1.53 million disk基于1530000盤數(shù)據(jù)的數(shù)據(jù)腐敗drives covering a time period of 41 months. We use the開蓋的時(shí)間期限為41個(gè)月。我們使用same data set as the one used in recent studies of latent在最近的研究中使用的相同的數(shù)據(jù)集sector errors 2 and disk failures 11. We identify the扇區(qū)錯(cuò)誤 2 和磁盤故障 11 。我們確定fraction

20、 of disks that develop corruption, examine factors that might affect the prevalence of corruption, such發(fā)展腐敗的磁盤組，檢查可能影響腐敗盛行的因素，例如as disk class and age, and study characteristics of corruption, such as spatial and temporal locality. To the best of作為磁盤類和年齡，研究腐敗的特征，如空間和時(shí)間的地方。到最好的our knowledge, this is t

21、he first study of silent data corruption in production and development systems.我們的知識(shí)，這是第一次在生產(chǎn)和發(fā)展系統(tǒng)中的無聲數(shù)據(jù)腐敗的研究。We classify data corruption into three categories based我們將數(shù)據(jù)分類為三類on how it is discovered: checksum mismatches, identity discrepancies, and parity incons它是如何發(fā)現(xiàn)：校驗(yàn)和不匹配，身份的差異，和奇偶incons（描述in det

22、ail in Section 2.3). We focus on checksum mismatches since they are found to occur the most. Our important observations include the following:在2.3節(jié)中詳細(xì)介紹。我們專注于校驗(yàn)和不匹配是因?yàn)樗麄儼l(fā)現(xiàn)發(fā)生的最。我們的重要意見包括以下內(nèi)容：(i) During the 41-month time period, we observe more（一）在41個(gè)月的時(shí)間內(nèi)，我們觀察到更多than 400, 000 instances of checksum mi

23、smatches, 8% of400，校驗(yàn)和不匹配的000個(gè)實(shí)例，8%which were discovered during RAID reconstruction, creating the possibility of real data loss. Even though the在空襲重建過程中發(fā)現(xiàn)的，創(chuàng)造了真實(shí)數(shù)據(jù)丟失的可能性。即使是rate of corruption is small, the discovery of checksum腐敗率小，校驗(yàn)和發(fā)現(xiàn)mismatches during reconstruction illustrates that data在重建過程中的不匹

24、配說明了數(shù)據(jù)corruption is a real problem that needs to be taken into腐敗是一個(gè)需要被納入的現(xiàn)實(shí)問題account by storage system designers.由存儲(chǔ)系統(tǒng)設(shè)計(jì)的帳戶。(ii) We find that nearline (SATA) disks and their adapters（ii）發(fā)現(xiàn)近線（SATA）磁盤和適配器develop checksum mismatches an order of magnitude開發(fā)一個(gè)量級(jí)的校驗(yàn)和不匹配more often than enterprise class (FC

25、) disks. Surprisingly,比企業(yè)級(jí)（足球）磁盤更經(jīng)常。令人驚訝的，enterprise class disks with checksum mismatches develop more of them than nearline disks with mismatches.校驗(yàn)和不匹配的企業(yè)級(jí)磁盤的發(fā)展超過了近線盤錯(cuò)位。(iii) Checksum mismatches are not independent occurrences both within a disk and within different disks in（iii）校驗(yàn)和不匹配的不獨(dú)立在磁盤和在不同的

26、磁盤上the same storage system.同一存儲(chǔ)系統(tǒng)。(iv) Checksum mismatches have tremendous spatial locality; on disks with multiple mismatches, it is often consecutive blocks that are affected.（四）校驗(yàn)和不匹配，有巨大的空間位置；對(duì)多錯(cuò)配盤，它往往是連續(xù)的數(shù)據(jù)塊的影響。(v) Identity discrepancies and parity inconsistencies do（五）身份差異和平價(jià)不一致occur, but affe

27、ct 3 to 10 times fewer disks than checksum發(fā)生，但影響3到10倍比較少的磁盤校驗(yàn)mismatches affect.錯(cuò)配影響。The rest of the paper is structured as follows. Section 2本文其余部分的結(jié)構(gòu)如下。第2節(jié)presents the overall architecture of the storage systems介紹存儲(chǔ)系統(tǒng)的總體架構(gòu)used for the study and Section 3 discusses the methodology used. Section 4 pr

28、esents the results of our analysis of checksum mismatches, and Section 5 presents the用于研究和3節(jié)討論所使用的方法。4節(jié)介紹了我國的校驗(yàn)和不匹配的分析結(jié)果，和5節(jié)介紹了results for identity discrepancies, and parity inconsistencies. Section 6 provides an anecdotal discussion of corruption, developing insights for corruption-proof storage結(jié)果的

29、身份差異，奇偶性不一致。第6節(jié)提供了一個(gè)軼事的腐敗問題，發(fā)展的見解，腐敗證據(jù)存儲(chǔ)system design. Section 7 presents related work and Section 8 provides a summary of the paper.系統(tǒng)設(shè)計(jì)。第7節(jié)介紹了有關(guān)工作和8節(jié)提供了一個(gè)總結(jié)的文件。2 Storage System Architecture2存儲(chǔ)系統(tǒng)架構(gòu)The data we analyze is from tens-of-thousands of production and development Network Appliance我們分析的數(shù)據(jù)來自

30、于成千上萬的生產(chǎn)和開發(fā)網(wǎng)絡(luò)設(shè)備TMTMstorage保管部systems (henceforth called the system) installed at hundreds of customer sites. This section describes the architecture of the system, its corruption detection mechanisms, and the classes of corruptions in our study.系統(tǒng)（此后稱為系統(tǒng)）安裝在數(shù)百個(gè)客戶網(wǎng)站。本節(jié)描述了該系統(tǒng)的體系結(jié)構(gòu)，其腐敗的檢測機(jī)制，并在研究腐敗類。2.1

31、 Storage Stack2.1存儲(chǔ)棧Physically, the system is composed of a storagecontroller that contains the CPU, memory, network interfaces, and storage adapters. The storage-controller物理上，該系統(tǒng)由包含CPU，內(nèi)存，一個(gè)storagecontroller網(wǎng)絡(luò)接口，存儲(chǔ)適配器。存儲(chǔ)控制器is connected to a set of disk shelves via Fibre Channel通過光纖通道連接到一組磁盤架上loops

32、. The disk shelves house individual disk drives.循環(huán)。磁盤架上的單個(gè)磁盤驅(qū)動(dòng)器。The disks may either be enterprise class FC disk drives磁盤可以是企業(yè)級(jí)的磁盤驅(qū)動(dòng)器or nearline serial ATA (SATA) disks. Nearline drives或近線串行ATA（SATA）硬盤。近線驅(qū)動(dòng)器use hardware adapters to convert the SATA interface to使用硬件適配器轉(zhuǎn)換為SATA接口the Fibre Channel proto

33、col. Thus, the storage-controller光纖通道協(xié)議。因此，存儲(chǔ)控制器views all drives as being Fibre Channel (however, for視圖所有驅(qū)動(dòng)器作為光纖通道（然而，對(duì)于the purposes of the study, we can still identify whether這項(xiàng)研究的目的，我們?nèi)匀豢梢源_定是否a drive is SATA and FC using its model type).硬盤是SATA和FC利用其模型類型）。The software stack on the storage-controll

34、er is composed of the WAFL在存儲(chǔ)控制器的軟件堆棧組成的細(xì)胞凋亡RRfile system, RAID, and storage文件系統(tǒng)，突襲和存儲(chǔ)layers. The file system processes client requests by issuing read and write operations to the RAID layer, which層。該文件系統(tǒng)處理客戶端請(qǐng)求，通過發(fā)布讀寫操作來處理層transforms the file system requests into logical disk block將文件系統(tǒng)請(qǐng)求轉(zhuǎn)換為邏輯磁盤塊re

35、quests and issues them to the storage layer. The RAID請(qǐng)求并將它們發(fā)布到存儲(chǔ)層?？找ulayer also generates parity for writes and reconstructs層也產(chǎn)生奇偶校驗(yàn)寫入和重構(gòu)data after failures. The storage layer is a set of customized device drivers that communicate with physical故障后的數(shù)據(jù)。存儲(chǔ)層是一組定制的設(shè)備驅(qū)動(dòng)程序，與物理通信disks using the SCSI command

36、set 23.使用SCSI命令集 23盤。2.2 Corruption Detection Mechanisms2.2腐敗檢測機(jī)制The system, like other commercial storage systems, is與其他商業(yè)存儲(chǔ)系統(tǒng)，該系統(tǒng)是designed to handle a wide range of disk-related errors.設(shè)計(jì)用于處理磁盤相關(guān)的廣泛錯(cuò)誤。The data integrity checks in place are designed to detect and recover from corruption errors so t

37、hat they are數(shù)據(jù)完整性檢查的目的是為了檢測和恢復(fù)從腐敗的錯(cuò)誤，使他們not propagated to the user. The system does not knowingly propagate corrupt data to the user under any circumstance.不傳播給用戶。在任何情況下，該系統(tǒng)不向用戶傳播腐敗數(shù)據(jù)。We focus on techniques used to detect silent data corruption, that is, corruptions not detected by the disk drive我們專

38、注于用來檢測靜默數(shù)據(jù)損壞，這是技術(shù)，通過硬盤檢測不到腐敗or any other hardware component. Therefore, we do not或任何其他硬件組件。因此，我們不describe techniques used for other errors, such as transport corruptions reported as SCSI transport errors or latent sector errors. Latent sector errors are caused by描述用于其他錯(cuò)誤的技術(shù)，如運(yùn)輸損壞報(bào)告為SCSI傳輸錯(cuò)誤或潛在扇區(qū)錯(cuò)誤。潛

39、在的部門錯(cuò)誤造成的physical problems within the disk drive, such as media磁盤驅(qū)動(dòng)器內(nèi)的物理問題，如媒體scratches, “high-fly” writes, etc. 2, 18, and detected by劃痕，“高飛”寫等 2，18 ，并檢測the disk drive itself by its inability to read or write sectors, or through its error-correction codes (ECC).磁盤驅(qū)動(dòng)器本身的讀寫扇區(qū)的無能，或通過其糾錯(cuò)碼（ECC）。In order

40、 to detect silent data corruptions, the system為了檢測沉默的數(shù)據(jù)損壞，系統(tǒng)stores extra information to disk blocks. It also periodically reads all disk blocks to perform data integrity存儲(chǔ)額外信息到磁盤塊。它還定期讀取所有磁盤塊來執(zhí)行數(shù)據(jù)完整性checks. We now describe these techniques in detail.支票。我們現(xiàn)在詳細(xì)描述這些技術(shù)。Corruption Class Possible Causes D

41、etection Mechanism Detection Operation腐敗類可能導(dǎo)致檢測機(jī)制的檢測操作Checksum mismatch Bit-level corruption; torn write; RAID block checksum Any disk read校驗(yàn)和錯(cuò)配位腐??；撕開寫；RAID塊校驗(yàn)磁盤讀misdirected write錯(cuò)誤的寫Identity discrepancy Lost or misdirected write File system-level block identity File system read身份差異丟失或誤導(dǎo)寫文件系統(tǒng)級(jí)的文件系統(tǒng)讀

42、取塊身份Parity inconsistency Memory corruption; lost write; RAID parity mismatch Data scrub奇偶性不一致的內(nèi)存損壞；丟失的寫；校驗(yàn)失配數(shù)據(jù)擦洗bad parity calculation差平價(jià)計(jì)算Table 1: Corruption classes summary.表1：腐敗類總結(jié)。(a) Format for enterprise class disks（一）企業(yè)級(jí)磁盤的格式520 520 520 520 520 520520 520 520 520 520 5204 KB4 KB文件系統(tǒng)數(shù)據(jù)塊520 520

43、520 52064byte Data64字節(jié)數(shù)據(jù)Integrity Segment完整性段(b) Format for nearline disks（b）為近線磁盤格式4 KB File system data block4 KB的文件系統(tǒng)數(shù)據(jù)塊512 512 512 512 512 512 512 512 512512 512 512 512 512 512 512 512 512448 bytes unused448字節(jié)未使用64byte Data64字節(jié)數(shù)據(jù)Integrity Segment +完整性段+(c) Structure of the data integrity segmen

44、t (DIS)（）數(shù)據(jù)完整性分部（解散）的結(jié)構(gòu).。Checksum of data block數(shù)據(jù)塊校驗(yàn)Identity of data block數(shù)據(jù)塊身份.Checksum of DIS校驗(yàn)和DISFigure 1: Data Integrity Segment. The figure shows the圖1：數(shù)據(jù)完整性段。圖顯示different on-disk formats used to store the data integrity segment of a disk block on (a) enterprise class drives with 520B用于存儲(chǔ)磁盤塊的數(shù)

45、據(jù)完整性段光盤格式的不同（一）與企業(yè)級(jí)硬盤520Bsectors, and on (b) nearline drives with 512B sectors. The figure also shows (c) the structure of the data integrity segment.部門，和（b）近線驅(qū)動(dòng)器512B扇區(qū)。圖還顯示了數(shù)據(jù)完整性段的結(jié)構(gòu)。In particular, in addition to the checksum and identity information, this structure also contains a checksum of itse

46、lf.特別是，除了校驗(yàn)和身份信息，該結(jié)構(gòu)還包含一個(gè)校驗(yàn)本身。2.2.1 Data Integrity Segment2.2.1數(shù)據(jù)完整段In order to detect disk block corruptions, the system為了檢測磁盤塊的損壞，系統(tǒng)writes a 64-byte data integrity segment along with each一個(gè)64字節(jié)的數(shù)據(jù)完整段以及每個(gè)disk block. Figure 1 shows two techniques for storing磁盤塊。圖1顯示了存儲(chǔ)的技術(shù)this extra information, and

47、also describes its structure.這個(gè)額外的信息，也描述了它的結(jié)構(gòu)。For enterprise class disks, the system uses 520-byte sectors. Thus, a 4-KB file system block is stored along with對(duì)于企業(yè)級(jí)磁盤，該系統(tǒng)使用520字節(jié)扇區(qū)。因此，一個(gè)4KB的文件系統(tǒng)的塊存儲(chǔ)在64 bytes of data integrity segment in eight 520-byte sectors. For nearline disks, the system uses the

48、default 512-byte sectors and store the data integrity segment for each八字節(jié)的數(shù)據(jù)完整性分部在520個(gè)64字節(jié)扇區(qū)。對(duì)于近線盤，系統(tǒng)將使用默認(rèn)的512字節(jié)扇區(qū)存儲(chǔ)數(shù)據(jù)完整性的一段set of eight sectors in the following sector. We find that在下列部門設(shè)置八個(gè)部門。我們發(fā)現(xiàn)the protection offered by the data integrity segment is數(shù)據(jù)完整性段所提供的保護(hù)well-worth the extra space needed t

49、o store them.很值得的額外空間來存儲(chǔ)它們。One component of the data integrity segment is a數(shù)據(jù)完整性段的一個(gè)組成部分是checksum of the entire 4 KB file system block. The整個(gè)4 KB的文件系統(tǒng)的塊校驗(yàn)。這個(gè)checksum is validated by the RAID layer whenever the校驗(yàn)和是由RAID層驗(yàn)證時(shí)data is read. Once a corruption has been detected, the數(shù)據(jù)讀取。一旦發(fā)現(xiàn)了腐敗，original bl

50、ock can usually be restored through RAID reconstruction. We refer to corruptions detected by RAIDlevel checksum validation as checksum mismatches.原始的塊通?？梢酝ㄟ^空襲重建恢復(fù)。我們指的raidlevel檢測校驗(yàn)和驗(yàn)證作為校驗(yàn)和錯(cuò)配的腐敗。A second component of the data integrity segment is數(shù)據(jù)完整性段的另一個(gè)組成部分是block identity information. In this case

51、, the fact that the塊身份信息。在這種情況下，事實(shí)上，file system is part of the storage system is utilized. The文件系統(tǒng)是利用存儲(chǔ)系統(tǒng)的一部分。這個(gè)identity is the disk blocks identity within the file system身份是文件系統(tǒng)中的磁盤塊的標(biāo)識(shí)(e.g., this block belongs to inode 5 at offset 100). This（例如，這一塊屬于inode 5偏移100）。這identity is cross-checked at file

52、 read time to ensure that在文件讀取時(shí)間時(shí)，要確保交叉檢查，以確保the block being read belongs to the file being accessed.被讀取的塊屬于被訪問的文件。If, on file read, the identity does not match, the data is如果，在文件讀取時(shí)，身份不匹配，數(shù)據(jù)是reconstructed from parity. We refer to corruptions that從奇偶校驗(yàn)。我們指的是腐敗，are not detected by checksums, but dete

53、cted through file沒有檢測到通過校驗(yàn)，但檢測到文件system identity validation as identity discrepancies.身份差異的系統(tǒng)身份驗(yàn)證。2.2.2 Data Scrubbing2.2.2數(shù)據(jù)清理In order to pro-actively detect errors, the RAID layer periodically scrubs all disks. A data scrub issues read operations for each physical disk block, computes a checksum o

54、ver its data, and compares the computed checksum to the checksum located in its data integrity segment. If the checksum comparison fails (i.e., a checksum為了積極檢測錯(cuò)誤，定期擦洗所有磁盤的RAID層。數(shù)據(jù)清洗問題讀操作的每個(gè)物理磁盤塊，計(jì)算校驗(yàn)和的數(shù)據(jù)，并比較計(jì)算的校驗(yàn)和校驗(yàn)和位于其完整的數(shù)據(jù)段。如果校驗(yàn)和比較失?。?，一個(gè)校驗(yàn)和mismatch), the data is reconstructed from other disks in

55、不匹配），數(shù)據(jù)從其他磁盤重建the RAID group, after those checksums are also verified.的RAID組，經(jīng)過校驗(yàn)和驗(yàn)證。If no reconstruction is necessary, the parity of the data如果沒有重建是必要的，數(shù)據(jù)的奇偶性blocks is generated and compared with the parity stored塊生成并與奇偶存儲(chǔ)in the parity block. If the parity does not match the verified data, the scru

56、b process fixes the parity by regenerating it from the data blocks. In a system protected by在奇偶校驗(yàn)塊。如果奇偶校驗(yàn)不匹配的驗(yàn)證數(shù)據(jù)，擦洗過程修復(fù)的奇偶性，通過再生的數(shù)據(jù)塊。在受保護(hù)的系統(tǒng)中double parity, it is possible to definitively tell which of雙奇偶校驗(yàn)，可以明確地告訴它the parity or data block is corrupt.奇偶或數(shù)據(jù)塊被損壞。We refer to these cases of mismatch between data and我們指的是這些情況下，數(shù)據(jù)之間的不匹配parity as parity inconsistencies. Note that data scrubs奇偶校驗(yàn)不一致。注意，數(shù)據(jù)服are unable to validate the extra file system identity information stored in the data integrity segment, since, by its