NetAppDataOntap核心基礎(chǔ)技術(shù)技術(shù)工程師培訓(xùn)ppt課件

1、1NetApp Confidential - Internal Use OnlyDataOntap中心技術(shù)Data OntapNetApp一致存儲(chǔ)架構(gòu)ApplicationOSSecurityVolume Mgmt (RAID)NTFS/UFSDisk Drivers通用操作系統(tǒng)中心技術(shù)：公用操作系統(tǒng)DataOntap70萬 行代碼，代碼精煉功能不多，但是專注于數(shù)據(jù)存儲(chǔ)，由于專所以高效對(duì)病毒免疫公用操作系統(tǒng)DataOntap5000 萬行代碼，代碼龐雜功能繁多，但是大部分和數(shù)據(jù)存儲(chǔ)無關(guān)，從而影響數(shù)據(jù)存儲(chǔ)性能開放系統(tǒng)，容易受病毒攻擊WAFL2022/7/12NetApp Confidential

2、 - Do Not Distribute4公用操作系統(tǒng)和文件系統(tǒng)-WAFL高效支持一切協(xié)議NetApp LUNs 是在Flex vol之內(nèi)建立的具備磁盤塊屬性的數(shù)據(jù)對(duì)象ted within FlexVol volumes.NetApp在SAN環(huán)境和NAS環(huán)境提供同樣的存儲(chǔ)效率經(jīng)過 NetApp的一致架構(gòu), 單一系統(tǒng)可以提供SAN和NAS功能，并且易于再配置和改動(dòng)應(yīng)對(duì)業(yè)務(wù)挑戰(zhàn)Disk DriveData ONTAP Logical ViewRAID GroupAggregateFlexVolLUNSATARAID-DPThin ProvisioningThin ReplicationSnapsh

3、ot CopiesVirtual CopiesDeduplicationAbstraction LayerEfficiency Feature存儲(chǔ)單元?jiǎng)澐諥ggr0Aggr1vol0vol1vol2vol3物理磁盤Aggregate(RAID組集合)FlexvolLUN物理層面邏輯層面存儲(chǔ)單元稱號(hào)與定義Aggregate每個(gè)控制器必需有獨(dú)立的Aggregate可以包含一個(gè)或多個(gè)RAID組VolumeNAS存儲(chǔ)單元虛擬化邏輯單元，可以動(dòng)態(tài)縮放20MB16TBWAFL文件系統(tǒng)構(gòu)造LUNSAN/IPsan 存儲(chǔ)單元虛擬化，可以動(dòng)態(tài)縮放數(shù)據(jù)格式由主機(jī)定義Data OntapNetApp一致存儲(chǔ)架構(gòu)Ne

4、tApp Raid 技術(shù)詳解1137222149 為什么其他廠商不提供RAID4支持，RAID4本身存在什么缺乏? Netapp如何處理RAID4本身的缺乏?3912213321112258812333+1+2+3=99-3-1-3=211512DDDDP不論是RAID4還是RAID5，都是基于單校驗(yàn)技術(shù)的RAID方法，只能防止因一塊硬盤缺點(diǎn)導(dǎo)致數(shù)據(jù)失效Netapp采用的RAID技術(shù)之一：RAID4RAID4有一塊公用的校驗(yàn)盤3521211813213713213891222DDDDDRAID5沒有公用的校驗(yàn)盤，一切的校驗(yàn)信息分散存儲(chǔ)在一切硬盤上用戶數(shù)據(jù)校驗(yàn)數(shù)據(jù)圖例此塊數(shù)據(jù)要從3修正為55+=

5、1122149修正為2修正為2修正為1修正為4RAID4：RAID組內(nèi)添加硬盤不需求進(jìn)展重構(gòu)9587P0000D3121113122133122DDDDRAID4有一塊公用的校驗(yàn)盤只需新添加的硬盤一切位都格式化為0，把硬盤添加到RAID4中就不會(huì)改動(dòng)各個(gè)條帶的校驗(yàn)值，因此整個(gè)RAID組不需求重構(gòu)RAID5：RAID組內(nèi)添加硬盤需求進(jìn)展重構(gòu)35212118132137132138DDDD91222DRAID5沒有公用的校驗(yàn)盤，一切的校驗(yàn)信息分散存儲(chǔ)在一切硬盤上00000D50827118為了堅(jiān)持RAID5分散存放校驗(yàn)信息的特點(diǎn)，新硬盤添加到RAID組后，整個(gè)RAID組需求進(jìn)展長時(shí)間的重構(gòu)，在這段

6、時(shí)間內(nèi)數(shù)據(jù)沒有RAID維護(hù)，且會(huì)嚴(yán)重影響性能D單校驗(yàn)RAID技術(shù)的RAID組重構(gòu)問題9587P3121113122133122DDDDRAID4有一塊公用的校驗(yàn)盤9 3 1 3 =25 1 1 1 =28 2 3 2 =17 1 1 2 =3RAID組的重構(gòu)，會(huì)給用戶帶來如下的影響： 需求較長的時(shí)間完成，且硬盤越大，所需的時(shí)間越長 會(huì)嚴(yán)重耗費(fèi)陣列的資源，特別是該RAID組內(nèi)的硬盤讀寫資源，因此會(huì)嚴(yán)重影響性能 在重構(gòu)期內(nèi)，整個(gè)RAID組沒有維護(hù)，假設(shè)再有一塊硬盤失效會(huì)導(dǎo)致用戶數(shù)據(jù)的喪失312111312213312295877121211DDDDPDP一個(gè)Raid組內(nèi)可以同時(shí)壞兩塊硬盤數(shù)據(jù)可靠性

7、比RAID-4 或 RAID-5 提高了 4000倍Netapp采用的RAID技術(shù)之二：RAID DP用戶數(shù)據(jù)程度交驗(yàn)數(shù)據(jù)對(duì)角交驗(yàn)數(shù)據(jù)圖例同樣的情況在 RAID-6形狀下140%5%10%15%20%Up to 5%Up to 2.6%Up to 17.9%FCATAFCATAFCATA17.9%1.7%2.6%.2%5%3%ATALess than 1 in a billion*介質(zhì)或磁道損壞 + 第二塊磁盤損壞的能夠性 雙校驗(yàn)(在16塊盤的RAIDDP磁盤組重構(gòu)期間)Protected withRAID-DP.0000000001%*規(guī)范的磁盤改換率 (每年)*磁盤驅(qū)動(dòng)器介質(zhì)或磁道損壞幾率

8、(300GB FC / 320GB SATA)*磁盤驅(qū)動(dòng)器介質(zhì)或磁道再有一塊盤損壞的幾率(在RAID4/5，8塊磁盤的磁盤組重構(gòu)的過程中)行業(yè)統(tǒng)計(jì): 磁盤驅(qū)動(dòng)器的改換率和磁盤介質(zhì)損壞的幾率隨著磁盤容量的增大在不斷的添加*資料來源: Network ApplianceData OntapNetApp一致存儲(chǔ)架構(gòu)NetApp Raid 技術(shù)詳解底層空間管理技術(shù)和性能優(yōu)化C底層存儲(chǔ)空間管理技術(shù)根ABD用戶數(shù)據(jù)此數(shù)據(jù)需求修正，從C修正為C空閑空間池CC其它廠商底層存儲(chǔ)空間管理技術(shù)在用戶數(shù)據(jù)修正時(shí)所采用的技術(shù)根ABD用戶數(shù)據(jù)空閑空間池在原來位置上直接利用新數(shù)據(jù)覆蓋老數(shù)據(jù)Netapp先進(jìn)的底層存儲(chǔ)空間管理

9、技術(shù)：WAFL在用戶數(shù)據(jù)修正時(shí)所采用的技術(shù)根ABD用戶數(shù)據(jù)空閑空間池CC1.從空閑空間池分配一個(gè)新的數(shù)據(jù)塊2.把修正以后的數(shù)據(jù)寫入到新分配的數(shù)據(jù)塊中3.修正指針，從指向老數(shù)據(jù)塊修正為指向新數(shù)據(jù)塊4.老數(shù)據(jù)塊數(shù)據(jù)不做任何修正，直接前往空閑空間池WAFL：所帶來的優(yōu)越性能硬盤硬盤硬盤硬盤硬盤硬盤硬盤硬盤硬盤硬盤硬盤校驗(yàn)盤Netapp WAFL的做法硬盤硬盤硬盤硬盤硬盤硬盤其它廠商的做法采用RAID5讀讀讀校驗(yàn)數(shù)據(jù)寫入用戶數(shù)據(jù)寫入讀硬盤硬盤硬盤硬盤硬盤校驗(yàn)盤其它廠商的做法采用RAID4讀讀讀讀用戶數(shù)據(jù)寫入校驗(yàn)數(shù)據(jù)寫入需求修正的數(shù)據(jù)修正以后的數(shù)據(jù)需求讀取的數(shù)據(jù)校驗(yàn)數(shù)據(jù)圖例 剛剛遺留的和RAID相關(guān)的兩

10、個(gè)問題為什么其他廠商不提供RAID4支持，RAID4本身存在什么缺乏? Netapp如何處理RAID4本身的缺乏? 會(huì)涉及到剛剛引見的單校驗(yàn)的RAID技術(shù)RAID4或RAID5如何進(jìn)展數(shù)據(jù)修正的技術(shù)其他廠商：RAID組的大小和性能成反比硬盤硬盤硬盤硬盤硬盤硬盤硬盤6D+1P的RAID5組：每個(gè)數(shù)據(jù)塊的修正都要涉及到其它五塊硬盤一樣位置的數(shù)據(jù)塊的讀取，然后進(jìn)展校驗(yàn)計(jì)算硬盤硬盤硬盤硬盤3D+1P的RAID5組：每個(gè)數(shù)據(jù)塊的修正只需涉及到其它兩塊硬盤一樣位置的數(shù)據(jù)塊的讀取，然后進(jìn)展校驗(yàn)計(jì)算需求修正的數(shù)據(jù)修正以后的數(shù)據(jù)需求讀取的數(shù)據(jù)校驗(yàn)數(shù)據(jù)圖例RAID組越小，性能越好，因此其他廠商普通都建議用來存放消

11、費(fèi)數(shù)據(jù)的RAID5的硬盤組采用3D+1P或4D+1P大小Netapp：RAID組的大小和性能成正比硬盤硬盤硬盤硬盤硬盤硬盤硬盤6D+1P的RAID4組：每6個(gè)數(shù)據(jù)塊的修正只需進(jìn)展一次校驗(yàn)計(jì)算，只會(huì)產(chǎn)生一個(gè)校驗(yàn)塊硬盤硬盤硬盤硬盤3D+1P的RAID4組：每3個(gè)數(shù)據(jù)塊的修正就要進(jìn)展一次校驗(yàn)計(jì)算，產(chǎn)生一個(gè)校驗(yàn)塊，這樣6個(gè)數(shù)據(jù)快的修正就要進(jìn)展兩次校驗(yàn)計(jì)算，產(chǎn)生兩個(gè)校驗(yàn)塊需求修正的數(shù)據(jù)修正以后的數(shù)據(jù)需求讀取的數(shù)據(jù)校驗(yàn)數(shù)據(jù)圖例RAID組越大，性能越好，因此Netapp普通建議用戶的RAID4硬盤組至少采用6D+1P的大小22DataDataDataDataParityParityDataWAFL 無需預(yù)先

12、分配空間(元數(shù)據(jù)和數(shù)據(jù)同樣對(duì)待，寫條帶條帶到最近的可用塊.條帶化到最近的可用塊減少磁盤尋道，到達(dá)最好的磁盤性能，防止熱點(diǎn)攀的產(chǎn)生!WAFL將隨機(jī)寫轉(zhuǎn)換成大的順序?qū)?WAFL 文件系統(tǒng)和 Data ONTAP OS結(jié)合高性能RAID-6 (RAID-DP)DDDDPPDDDDPPDDDDPPDDDDPP4K blocks,parity calculatedin memoryIncoming WritesNVRAM傳統(tǒng)的文件系統(tǒng)NTFS, UFS, FAT32等,將數(shù)據(jù)寫到預(yù)先分配好的位置，導(dǎo)致“熱點(diǎn)校驗(yàn)磁盤DataDRAMData OntapNetApp一致存儲(chǔ)架構(gòu)NetApp Raid 技術(shù)詳

13、解底層空間管理技術(shù)和性能優(yōu)化NetApp快照 恢復(fù)和克隆快照數(shù)據(jù)消費(fèi)數(shù)據(jù)有數(shù)據(jù)快照的情況沒有數(shù)據(jù)快照的情況Netapp數(shù)據(jù)快照技術(shù)：SnapshotABCD空閑空間池消費(fèi)數(shù)據(jù)ABC空閑空間池DDD用戶最關(guān)懷的和數(shù)據(jù)快照相關(guān)的兩個(gè)問題：數(shù)據(jù)快照對(duì)性能的影響程度數(shù)據(jù)快照所需耗費(fèi)的存儲(chǔ)空間快照數(shù)據(jù)消費(fèi)數(shù)據(jù)有數(shù)據(jù)快照的情況沒有數(shù)據(jù)快照的情況其它廠商的數(shù)據(jù)快照技術(shù)ABCD用戶數(shù)據(jù)消費(fèi)數(shù)據(jù)ABC用戶數(shù)據(jù)數(shù)據(jù)快照公用存儲(chǔ)空間DDDD不同的數(shù)據(jù)快照實(shí)現(xiàn)的性能差別CX500 performance with their snapshots goes to 6TBNetapp的FlexVol技術(shù)提高了存儲(chǔ)空間的運(yùn)

14、用率80%利用率Reclaimed SpaceAvailable for GrowthFlexVol提升 I/O 性能原有的架構(gòu)卷的性能被它的磁盤數(shù)量限制“熱點(diǎn)卷不能從其他卷的磁盤上獲得協(xié)助NetApp FlexVol的技術(shù)各個(gè)卷可以共享所以磁盤的聚合性能Netapp的Flexvol技術(shù)所采用的先進(jìn)的存儲(chǔ)空間組織方式RG1RG2RG3磁盤LUN圖例運(yùn)用主機(jī)1運(yùn)用主機(jī)N一切硬盤先分成相應(yīng)的RAID組一切的RAID組組成存儲(chǔ)池在存儲(chǔ)池上面創(chuàng)建LUNNetapp的Flexvol技術(shù)完全消除了訪問熱點(diǎn)的產(chǎn)生，充分發(fā)揚(yáng)了存儲(chǔ)系統(tǒng)的潛在性能RG1RG2RG3磁盤LUN圖例主機(jī)IO運(yùn)用主機(jī)1運(yùn)用主機(jī)N主機(jī)一

15、切的IO都平均分散在陣列一切的硬盤上，徹底消除了陣列訪問熱點(diǎn)問題軟件產(chǎn)品和功能引見 Netapp創(chuàng)新的RAID技術(shù) 公用操作系統(tǒng)(Dataontap)和底層存儲(chǔ)空間管理技術(shù)(WAFL) 先進(jìn)的數(shù)據(jù)快照(Snapshot)和快速恢復(fù)(Snaprestore)技術(shù) 先進(jìn)的存儲(chǔ)空間分配和組織技術(shù)(Flexvol) 先進(jìn)的數(shù)據(jù)克隆技術(shù)(Flexclone) 運(yùn)用系統(tǒng)存儲(chǔ)訪問優(yōu)先級(jí)控制技術(shù)(Flexshare) 先進(jìn)的基于磁盤陣列的數(shù)據(jù)容災(zāi)技術(shù)(Snapmirror)只需中心運(yùn)用集中了中心和邊緣運(yùn)用不區(qū)分運(yùn)用等級(jí)集中了中心和邊緣運(yùn)用利用Flexshare區(qū)分運(yùn)用等級(jí)高優(yōu)先級(jí)高優(yōu)先級(jí)低優(yōu)先級(jí)低優(yōu)先級(jí)Fle

16、xshare：保證中心運(yùn)用的存儲(chǔ)訪問性能存儲(chǔ)網(wǎng)絡(luò)存儲(chǔ)網(wǎng)絡(luò)存儲(chǔ)網(wǎng)絡(luò)軟件產(chǎn)品和功能引見 Netapp創(chuàng)新的RAID技術(shù) 公用操作系統(tǒng)(Dataontap)和底層存儲(chǔ)空間管理技術(shù)(WAFL) 先進(jìn)的數(shù)據(jù)快照(Snapshot)和快速恢復(fù)(Snaprestore)技術(shù) 先進(jìn)的存儲(chǔ)空間分配和組織技術(shù)(Flexvol) 先進(jìn)的數(shù)據(jù)克隆技術(shù)(Flexclone) 運(yùn)用系統(tǒng)存儲(chǔ)訪問優(yōu)先級(jí)控制技術(shù)(Flexshare) 先進(jìn)的基于磁盤陣列的數(shù)據(jù)容災(zāi)技術(shù)(Snapmirror)50SnapMirror特點(diǎn)- 不同協(xié)議數(shù)據(jù),一致冗災(zāi)架構(gòu)NAS消費(fèi)系統(tǒng)ISCSISAN冗災(zāi)系統(tǒng)SnapMirrorSnapMirrorS

17、napMirrorNASISCSISAN51SnapMirror ApplicationsSnapMirror 技術(shù)將數(shù)據(jù)鏡像到一個(gè)或多個(gè)FAS存儲(chǔ)系統(tǒng)上。基于存儲(chǔ)設(shè)備底層的增量數(shù)據(jù)復(fù)制, 對(duì)主機(jī)效力器上層運(yùn)用和數(shù)據(jù)庫無影響。支持多種傳輸方式FC 和 IP 。異地備份數(shù)據(jù)拷貝提供只讀功能和缺點(diǎn)切換訪問?？蛻暨x擇適當(dāng)?shù)耐郊?jí)別同步、準(zhǔn)同步和異步，從而實(shí)現(xiàn)了較高的災(zāi)難恢復(fù)維護(hù)級(jí)別。完全基于IP網(wǎng)絡(luò)，無需公用光纖鏈路及多路波分設(shè)備，沒有傳輸間隔限制。運(yùn)用異步方式，對(duì)主機(jī)系統(tǒng)效力器I/O性能無影響。SnapMirror支持多種拓?fù)浞绞降膹?fù)制，包括：1對(duì)多；多對(duì)1；級(jí)聯(lián)復(fù)制等。SnapMirror支持對(duì)

18、多種數(shù)據(jù)的復(fù)制，包括：基于卷的復(fù)制，基于qtree的復(fù)制，支持對(duì)NAS / SAN /iSCSI中的數(shù)據(jù)復(fù)制。SnapMirror只傳輸變化的數(shù)據(jù)塊，可以大大節(jié)省網(wǎng)絡(luò)帶寬和減輕FAS存儲(chǔ)系統(tǒng)的負(fù)載。 52SnapMirror ModesSynchronous SnapMirror34Zero data lossDistance limitedPerformance impact142Small data lossNo distance limitNo performance impactSemi-Synchronous SnapMirror12Data loss 1 minute No dis

19、tance limitNo performance impactAsynchronous SnapMirror3Every Write2Every WriteAB312Changed blocksSet intervals153Snap ASnapMirror operation(異步方式)Baseline TransferSource File SystemRemote File System54Snap ASnapMirror operation (異步方式)Baseline TransferSource File SystemRemote File SystemSource file s

20、ystem continues to change during transferRemote file system is now consistent, and a mirror of the Snapshot A file systemCompletedSnapMirror to tape for seeding mirrored pairs55Snap ASnapMirror operation (異步方式)Incremental TransferSource File SystemRemote File SystemSnap BRemote file system is now co

21、nsistent, and a mirror of the Snapshot B file systemCompleted56SnapMirror operation (異步方式)Source File SystemRemote File SystemSnap BSnap CIncremental TransferCompletedRemote file system is now consistent, and a mirror of the Snap C file systemThe NetApp Model57NetApp Confidential - Internal Use Only

22、DepartmentalEnterpriseEnterpriseSANNASDepartmentalUnified StorageLANiSCSIFibreChannelEthernetScaling Up and Out with Data ONTAP 858NetApp Confidential - Internal Use OnlyDepartmentalEnterpriseEnterpriseSANNASDepartmentalLANiSCSIFibreChannelEthernetData ONTAP 8.0 NetApp主要磁盤存儲(chǔ)產(chǎn)品家族59NetApp Confidential

23、 - Internal Use OnlyFAS2xxx68 - 432TB68 - 144drivesSAS, FC & SATAFAS62xx2400 - 2880TB1200 - 1440 drivesFC & SAS & SATA480 1920TB240 - 960 drivesFC & SAS & SATAAll systems runData ONTAPFAS32xx異構(gòu)存儲(chǔ)設(shè)備支持480TB 1920TB240 - 960 LUNsV32xx1200 2880TB1200 - 1440 LUNsV62xx Small Medium Large X-LargeStorage Gri

24、d336 - 14112TB336 - 14112 drivesFC & SATANetApp一致存儲(chǔ)架構(gòu)60FCiSCSIFCoENFSCIFS對(duì)象協(xié)議的一致存儲(chǔ)架構(gòu)的一致Cluster Cloud ModeRAID-DP FlexVolEnd to end DeduplicationSnapShot Smart copiesSnapLock ComplianceSnapVaultSnapMirrorMetroCluster可靠性100km一致的軟件功能一致的運(yùn)用集成SnapManager Application Integration 閃存FlexCachepNFS1234NetApp一致

25、存儲(chǔ)架構(gòu)61一樣的工具和流程：學(xué)習(xí)一次，處處適用集成數(shù)據(jù)管理集成的數(shù)據(jù)維護(hù)供應(yīng)商之間采用一致規(guī)范企業(yè)級(jí)一致存儲(chǔ)平臺(tái)一致管理562NetApp Confidential - Do Not DistributeNetAppData ONTAPFCoE,FC,IP, & NAS一致的體系架構(gòu)設(shè)計(jì)一種管理界面完全的互操作學(xué)習(xí)一種、掌握全部FAS2040FAS2240FAS3210FAS3240FAS6280FAS6210FAS6240FAS3270跨平臺(tái)無縫晉級(jí)真正一致最高效率極其靈敏2400TB1200 個(gè)驅(qū)動(dòng)器3TB FlashCache1920TB960 個(gè)驅(qū)動(dòng)器2TB FlashCache12

26、00TB600 個(gè)驅(qū)動(dòng)器1TB FlashCache2880TB1440 個(gè)驅(qū)動(dòng)器6TB FlashCache2880TB1440 驅(qū)動(dòng)器8TB FlashCache480TB 240 個(gè)驅(qū)動(dòng)器 512GB FlashCacheData ONTAP 8.1 存儲(chǔ)集群A1C1LUNLUN B2ABA1 A3R CLUN B1C1LUNLUNAA2按需擴(kuò)容和負(fù)載平衡化數(shù)據(jù)層次化存儲(chǔ)高性能存儲(chǔ)廉價(jià)存儲(chǔ)C1LUNLUN B2ABA1 A3R CLUNA2 B1硬件在線晉級(jí) B2ABA1 A3R CLUNA2 B1C1LUNLUNAlways on availabilityOn demand flexi

27、bilityOperational efficiency64NetApp Confidential - Internal Use Only65NetApp Confidential - Internal Use OnlyNetApp Snapshot TechnologyInstant zero-space copyNetApp WAFL writes all data to available free space without writing over previous dataCreating a Snapshot copy involves merely retaining poin

28、ters to original data blocksPerformance impact is near zero because no additional disk operations are required beyond normal write behaviorSpace is efficient because only new data written takes up incremental space on diskScheduled or manual, up to 254 snapshots per volumeReadily accessible via a sp

29、ecial subdirectoryNetApp Storage SystemSnapshotCopiesNetApp Snapshot TechnologyTake snapshot 1Copy pointers onlyNo data movement66NetApp Confidential - Internal Use OnlyABCABCSnap 1Blocks in LUN or FileBlocks on the DiskABCNetApp Snapshot Technology67NetApp Confidential - Internal Use OnlyABCSnap 1B

30、locks on the DiskABCABCB1B1Take snapshot 1Continue writing dataWrite data anywhereBlocks in LUN or FileNetApp Snapshot Technology68NetApp Confidential - Internal Use OnlyAB1CABCSnap 1Blocks on the DiskBABCACB1B1Snap 2ABCSnap 1Take snapshot 1Continue writing dataTake snapshot 2Copy pointers onlyNo da

31、ta movementBlocks in LUN or FileNetApp Snapshot TechnologyTake snapshot 169NetApp Confidential - Internal Use OnlyAB1CSnap 2ABCSnap 1Blocks on the DiskABCABCContinue writing dataTake snapshot 2Continue writing dataWrite data anywhereB1B1C2C2Blocks in LUN or FileNetApp Snapshot TechnologyTake snapsho

32、t 170NetApp Confidential - Internal Use OnlyAB1C2Snap 3AB1CSnap 2Blocks on the DiskABCABCContinue writing dataTake snapshot 2Continue writing dataTake snapshot 3Simplicity of model =Best disk utilizationFastest performanceUnlimited snapshotsB1B1C2C2ABCSnap 1Blocks in LUN or FileEveryone Elses Snapsh

33、ots71NetApp Confidential - Internal Use OnlyOthers SnapshotsTake snapshot 1:Create copy out region 172NetApp Confidential - Internal Use OnlyBlocks on the DiskABCABCCreate pointers to old blocks and copy outCopyOut 1ABCSnap 1Blocks in LUNOthers SnapshotsTake snapshot 173NetApp Confidential - Interna

34、l Use OnlyABCSnap 1Blocks on the DiskABCACB1Read old block; write to copy out regionCopyOut 1ABCSnap 1Continue writing dataBlock changesUpdate snap pointer to copy out regionUpdate block on diskOne write requires:1 read (old data)1 write (old data)1 write (new data)3x overheadBBB1Blocks in LUNOthers SnapshotsTake snapshot 174NetApp Confidential - Internal Use OnlyBlocks on the DiskABCACB1CopyOut 1ABCSnap 1ABCSnap 1copyout 1Continue writing dataTake snapshot 2:Create copy out region 2CopyOut