RAID工作原理图解及特点分析Word下载.docx

RAID工作原理图解及特点分析Word下载.docx

《RAID工作原理图解及特点分析Word下载.docx》由会员分享，可在线阅读，更多相关《RAID工作原理图解及特点分析Word下载.docx（9页珍藏版）》请在冰豆网上搜索。

2.Thefailureofjustonedrivewillresultinalldatainanarraybeinglost

3.Shouldneverbeusedinmissioncriticalenvironments

RecommendedApplications

1.VideoProductionandEditing

2.ImageEditing

3.Pre-PressApplications

4.Anyapplicationrequiringhighbandwidth

RAIDLevel0requiresaminimumof2drivestoimplement

RAID1工作原理图解及特点分析

RAID1:

MirroringandDuplexing

1.OneWriteortwoReadspossiblepermirroredpair

2.TwicetheReadtransactionrateofsingledisks,sameWritetransactionrateassingledisks

3.100%redundancyofdatameansnorebuildisnecessaryincaseofadiskfailure,justacopytothereplacementdisk

4.Transferrateperblockisequaltothatofasingledisk

5.Undercertaincircumstances,RAID1cansustainmultiplesimultaneousdrivefailures

6.SimplestRAIDstoragesubsystemdesign

1.HighestdiskoverheadofallRAIDtypes（100%）-inefficient

2.TypicallytheRAIDfunctionisdonebysystemsoftware,loadingtheCPU/Serverandpossiblydegradingthroughputathighactivitylevels.Hardwareimplementationisstronglyrecommended

3.Maynotsupporthotswapoffaileddiskwhenimplementedin“software”

1.Accounting

2.Payroll

3.Financial

4.Anyapplicationrequiringveryhighavailability

ForHighestperformance,thecontrollermustbeabletoperformtwoconcurrentseparateReadspermirroredpairortwoduplicateWritespermirroredpair.

RAIDLevel1requiresaminimumof2drivestoimplement

RAID2工作原理图解及特点分析

RAID2:

HammingCodeECC

1.“Onthefly”dataerrorcorrection

2.Extremelyhighdatatransferratespossible

3.Thehigherthedatatransferraterequired,thebettertheratioofdatadiskstoECCdisks

4.RelativelysimplecontrollerdesigncomparedtoRAIDlevels3,4&

5

1.VeryhighratioofECCdiskstodatadiskswithsmallerwordsizes-inefficient

2.Entrylevelcostveryhigh-requiresveryhightransferraterequirementtojustify

3.Transactionrateisequaltothatofasinglediskatbest（withspindlesynchronization）

4.Nocommercialimplementationsexist/notcommerciallyviable

Eachbitofdatawordiswrittentoadatadiskdrive（4inthisexample:

0to3）.EachdatawordhasitsHammingCodeECCwordrecordedontheECCdisks.OnRead,theECCcodeverifiescorrectdataorcorrectssinglediskerrors.

RAID3工作原理图解及特点分析

RAID3:

Paralleltransferwithparity

1.VeryhighReaddatatransferrate

2.VeryhighWritedatatransferrate

3.Diskfailurehasaninsignificantimpactonthroughput

4.LowratioofECC（Parity）diskstodatadisksmeanshighefficiency

1.Transactionrateequaltothatofasinglediskdriveatbest（ifspindlesaresynchronized）

2.Controllerdesignisfairlycomplex

3.Verydifficultandresourceintensivetodoasa“software”RAID

1.VideoProductionandlivestreaming

3.VideoEditing

4.PrepressApplications

5.Anyapplicationrequiringhighthroughput

Thedatablockissubdivided（”striped”）andwrittenonthedatadisks.StripeparityisgeneratedonWrites,recordedontheparitydiskandcheckedonReads.

RAIDLevel3requiresaminimumof3drivestoimplement

RAID4工作原理图解及特点分析

RAID4:

IndependentDatadiskswithsharedParitydisk

1.VeryhighReaddatatransactionrate

2.LowratioofECC（Parity）diskstodatadisksmeanshighefficiency

3.HighaggregateReadtransferrate

1.Quitecomplexcontrollerdesign

2.WorstWritetransactionrateandWriteaggregatetransferrate

3.Difficultandinefficientdatarebuildintheeventofdiskfailure

4.BlockReadtransferrateequaltothatofasingledisk

Eachentireblockiswrittenontoadatadisk.ParityforsamerankblocksisgeneratedonWrites,recordedontheparitydiskandcheckedonReads.

RAIDLevel4requiresaminimumof3drivestoimplement

RAID5工作原理图解及特点分析

RAID5:

IndependentDatadiskswithdistributedparityblocks

1.HighestReaddatatransactionrate

2.MediumWritedatatransactionrate

3.LowratioofECC（Parity）diskstodatadisksmeanshighefficiency

4.Goodaggregatetransferrate

1.Diskfailurehasamediumimpactonthroughput

2.Mostcomplexcontrollerdesign

3.Difficulttorebuildintheeventofadiskfailure（ascomparedtoRAIDlevel1）

4.Individualblockdatatransferratesameassingledisk

1.FileandApplicationservers

2.Databaseservers

3.WWW,E-mail,andNewsservers

4.Intranetservers

5.MostversatileRAIDlevel

Eachentiredatablockiswrittenonadatadisk;

parityforblocksinthesamerankisgeneratedonWrites,recordedinadistributedlocationandcheckedonReads.

RAIDLevel5requiresaminimumof3drivestoimplement

RAID6工作原理图解及特点分析

RAID6:

IndependentDatadiskswithtwoindependentdistributedparityschemes

1.RAID6isessentiallyanextensionofRAIDlevel5whichallowsforadditionalfaulttolerancebyusingasecondindependentdistributedparityscheme（two-dimensionalparity）

2.Dataisstripedonablocklevelacrossasetofdrives,justlikeinRAID5,andasecondsetofparityiscalculatedandwrittenacrossallthedrives;

RAID6providesforanextremelyhighdatafaulttoleranceandcansustainmultiplesimultaneousdrivefailures

3.Perfectsolutionformissioncriticalapplications

1.Verycomplexcontrollerdesign

2.Controlleroverheadtocomputeparityaddressesisextremelyhigh

3.Verypoorwriteperformance

4.RequiresN+2drivestoimplementbecauseoftwo-dimensionalparityscheme

RAID7工作原理图解及特点分析

RAID7:

OptimizedAsynchronyforHighI/ORatesaswellasHighDataTransferRates

ArchitecturalFeatures:

1.AllI/Otransfersareasynchronous,independentlycontrolledandcachedincludinghostinterfacetransfers

2.Allreadsandwritearecentrallycachedviathehighspeedx-bus

3.Dedicatedparitydrivecanbeonanychannel

4.Fullyimplementedprocessorientedrealtimeoperatingsystemresidentonembeddedarraycontrolmicroprocessor

5.Embeddedrealtimeoperatingsystemcontrolledcommunicationschannel

6.OpensystemusesstandardSCSIdrives,standardPCbuses,motherboardsandmemorySIMMs

7.Highspeedinternalcachedatatransferbus（X-bus）

8.Paritygenerationintegratedintocache

9.Multipleattacheddrivedevicescanbedeclaredhotstandbys

10.Manageability:

SNMPagentallowsforremotemonitoringandmanagement

1.Overallwriteperformanceis25%to90%betterthansinglespindleperformanceand1.5to6timesbetterthanotherarraylevels

2.Hostinterfacesarescalableforconnectivityorincreasedhosttransferbandwidth

3.Smallreadsinmultiuserenvironmenthaveveryhighcachehitrateresultinginnearzeroaccesstimes

4.Writeperformanceimproveswithanincreaseinthenumberofdrivesinthearray

5.Accesstimesdecreasewitheachincreaseinthenumberofactuatorsinthearray

6.Noextradatatransfersrequiredforparitymanipulation

7.RAID7isaregisteredtrademarkofStorageComputerCorporation.

1.Onevendorproprietarysolution

2.ExtremelyhighcostperMB

3.Veryshortwarranty

4.Notuserserviceable

5.PowersupplymustbeUPStopreventlossofcachedata

RAID10工作原理图解及特点分析

工作原理,图解,RAID

RAID10:

VeryHighReliabilitycombinedwithHighPerformance

1.RAID10isimplementedasastripedarraywhosesegmentsareRAID1arrays

2.RAID10hasthesamefaulttoleranceasRAIDlevel1

3.RAID10hasthesameoverheadforfault-toleranceasmirroringalone

4.HighI/OratesareachievedbystripingRAID1segments

5.Undercertaincircumstances,RAID10arraycansustainmultiplesimultaneousdrivefailures

6.ExcellentsolutionforsitesthatwouldhaveotherwisegonewithRAID1butneedsomeadditionalperformanceboost

7.Veryexpensive/Highoverhead

8.Alldrivesmustmoveinparalleltopropertrackloweringsustainedperformance

9.Verylimitedscalabilityataveryhighinherentcost

Databaseserverrequiringhighperformanceandfaulttolerance

RAIDLevel10requiresaminimumof4drivestoimplement