archive and records managementâ€”fiscal year 2010 offline archive

U.S. Department of the InteriorU.S. Geological Survey

Open-File Report 2010–1222

Archive and Records Management—Fiscal Year 2010 Offline Archive Media Trade Study


By Tom Bodoh, Ken Boettcher, Ken Gacke, Cheryl Greenhagen, and Al Engelbrecht

Open-File Report 2010–1222

U.S. Department of the InteriorU.S. Geological Survey

U.S. Department of the InteriorKEN SALAZAR, Secretary

U.S. Geological SurveyMarcia K. McNutt, Director

U.S. Geological Survey, Reston, Virginia: 2010

For more information on the USGS—the Federal source for science about the Earth, its natural and living resources, natural hazards, and the environment, visit http://www.usgs.gov or call 1-888-ASK-USGS

For an overview of USGS information products, including maps, imagery, and publications, visit http://www.usgs.gov/pubprod

To order this and other USGS information products, visit http://store.usgs.gov

Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Although this report is in the public domain, permission must be secured from the individual copyright owners to reproduce any copyrighted materials contained within this report.

Suggested citation:Bodoh, Tom, Boettcher, Ken, Gacke, Ken, Greenhagen, Cheryl, and Engelbrecht, Al, 2010, Archive and records manage-ment—Fiscal year 2010 offline archive media trade study: U.S. Geological Survey Open-File Report 2010–1222, 15 p. plus app.

iii

Preface

This document contains the Offline Archive Media Trade Study prepared by Stinger Ghaffarian Technologies, Inc. (SGT) for the U.S. Geological Survey (USGS). This trade study presents the background, technical assessment, test results, and recommendations.

v

Contents

Preface ...........................................................................................................................................................iiiAbstract ...........................................................................................................................................................1Revision History..............................................................................................................................................1

February 2004.........................................................................................................................................1September 2006 .....................................................................................................................................1June 2008................................................................................................................................................1June 2010................................................................................................................................................2

Introduction.....................................................................................................................................................2Purpose and Scope ..............................................................................................................................2Background............................................................................................................................................2Data Integrity .........................................................................................................................................2Selection Criteria ..................................................................................................................................3Dismissed Technologies ......................................................................................................................4

Magnetic Disk...............................................................................................................................4Solid State Disk (SSD) .................................................................................................................4CD-ROM, DLT 8000, QIC, Mammoth, and Erasable Optical (EO) ..........................................4Oracle 9840....................................................................................................................................4Quantum DLT.................................................................................................................................4Tandberg/Exabyte VXA320, Sony SAIT-1/SAIT-2 ....................................................................4DVD, HD-DVD, Blu-Ray ...............................................................................................................5Newer Optical Technologies ......................................................................................................5

Technical Assessment ..................................................................................................................................5Analysis ..................................................................................................................................................5Oracle T10000B......................................................................................................................................6

Advantages ...................................................................................................................................6Disadvantages ..............................................................................................................................6Summary........................................................................................................................................6

Oracle T10000C ......................................................................................................................................7Advantages ...................................................................................................................................7Disadvantages ..............................................................................................................................7Summary........................................................................................................................................7

HP LTO5...................................................................................................................................................7Advantages ...................................................................................................................................7Disadvantages ..............................................................................................................................7Summary........................................................................................................................................8

HP LTO6...................................................................................................................................................8Advantages ...................................................................................................................................8Disadvantages ..............................................................................................................................9Summary........................................................................................................................................9

IBM TS1130 ............................................................................................................................................9Advantages ...................................................................................................................................9Disadvantages ..............................................................................................................................9Summary........................................................................................................................................9

vi

IBM TS1140 ..........................................................................................................................................10Advantages .................................................................................................................................10Disadvantages ............................................................................................................................10Summary......................................................................................................................................10

Tables .............................................................................................................................................................10Design Criteria .....................................................................................................................................10Transfer Rate .......................................................................................................................................10Capacity ................................................................................................................................................11Cost Analysis .......................................................................................................................................113.5 Scenarios .......................................................................................................................................12Vendor Analyses .................................................................................................................................12Drive Compatibility ..............................................................................................................................12Ranking Summary ...............................................................................................................................13

Conclusions and Recommendations for USGS Offline Archiving Requirements ..............................13Weighted Decision Matrix .................................................................................................................13Conclusions and Notes ......................................................................................................................14Recommendations ..............................................................................................................................14

Appendix: Citations ......................................................................................................................................17Vendor Sites .........................................................................................................................................17Consortium Sites .................................................................................................................................17Other......................................................................................................................................................17

Figures 1. Screen capture showing the Oracle roadmap (uncompressed) ..........................................6 2. Screen capture showing the LTO roadmap (with 2:1 compression) ....................................8 3. Screen capture showing the IBM roadmap (uncompressed) ..............................................9

Tables 1. Recent and current archive technologies used at the U.S. Geological Survey .................2 2. Tape drive markets and characteristics ...................................................................................3 3. Technology comparison ..............................................................................................................5 4. Design criteria and target market ...........................................................................................10 5. Transfer rates ..............................................................................................................................11 6. Storage capacities .....................................................................................................................11 8. Scenario costs (drives, media) .................................................................................................12 7. Drive and media costs ...............................................................................................................12 9. Vendor analyses .........................................................................................................................12 10. Drive compatibility ......................................................................................................................13 11. Ranking summaries ....................................................................................................................13

vii

Abbreviations and AcronymsAIT AdvancedIntelligentTapeBER BitErrorRateCD CompactDiscCD-ROM CompactDisc-ReadOnlyMemoryCERN ConseilEuropeanpourlaRecherchéNucleaireCRC CyclicRedundancyCheckCPU CentralProcessingUnitDCT DigitalCassettetapeDLT DigitalLinearTapeDVD DigitalVideoDiscEO ErasableOpticalEOT EndoftapeEROS EarthResourcesObservationandScienceFYyy FiscalYearyyGB Gigabytes(1,024MB,or1,073,741,824bytes)Gbit/sec GigabitpersecondHD-DVD HighDefinitionDigitalVersatileDisc(formerlyDigitalVideoDisc)HDT HighDensityTapeHP Hewlett-PackardHVD HolographicVersatileDiscHW HardwareIBM InternationalBusinessMachinesLPDAAC LandProcessesDistributedActiveArchiveCenterLTO LinearTape-OpenMB Megabytes(1,048,576bytes)MB/sec MegabytespersecondNARA NationalArchivesandRecordsAdministrationQIC Quarter-inchCartridgeSAIT SuperAdvancedIntelligentTapeSDLT SuperDigitalLinearTapeSGT StingerGhaffarianTechnologies,Inc.SSD SolidStateDiskSTK StorageTek(subsequentlyboughtbySun,whichwasboughtbyOracle)TB Terabytes(1,024GBor1,099,511,627,776bytes)TBD ToBeDecided/DeterminedUSGS U.S.GeologicalSurvey


By Tom Bodoh1, Ken Boettcher1, Ken Gacke1, Cheryl Greenhagen1, and Al Engelbrecht1

1StingerGhaffarianTechnologies,Inc.,contractortotheU.S.GeologicalSurvey,workperformedunderUSGScontract08HQCN0005.

AbstractThisdocumentisatradestudycomparingofflinedigital

archivestoragetechnologies.Thedocumentcomparesandassessesseveraltechnologiesandrecommendswhichtech-nologiescouldbedeployedasthenextgenerationstandardfortheU.S.GeologicalSurvey(USGS).Archivesmustregularlymigratetothenextgenerationofdigitalarchivetechnology,andthetechnologyselectedmustmaintaindataintegrityuntilthenextmigration.Thisdocumentisthefiscalyear2010(FY10)revisionofastudycompletedinFY01andrevisedinFY03,FY04,FY06,andFY08.

Revision History

February 2004

• Addedrevisionhistorypage.NorevisionhistoryisavailablefortheFY03revision.

• Changedtoallowforconsiderationofhelicalscanaslongascertainperformancecriteriaaremet.

• AddedLinearTape-Open(LTO)2asacurrentarchivetechnology.

• AddedSuperAdvancedIntelligentTape(SAIT)-1andSuperDigitalLinearTape(SDLT)600asconsidereddrives.

• ReplacedInternationalBusinessMachines(IBM)3590withIBM3592.

• RemovedLTO1andSDLT320fromthestudy.

• Consideredalldrivesinthestudy.

• Increasedtheminimumspecificationsforcapacityandtransferrate.

• Reworkedcostscenarios,andreducedthenumberofcostscenariostothree.

• Removedtransfertimescenarios.

• Removedmaintenancefromcostscenarios.

• Removedcriteriashowingmulti-vendoravailabilityasanadvantage.

September 2006

• Overallrefreshofstudy.

• Reviseddescriptionofdriveclasses(enterprise,backup).

• AddedLTO3,TS1120,T10000,andDLT-S4ascurrenttechnologiesandremoveddrivestheyreplaced.

• AddedLTO4andSAIT-2asfuturetechnologies.

• Madevendoranalysesformulamoreequitablebyincreasingweightingofcompanyage.

• Addedcitationappendix.

June 2008

• Overallrefreshofstudy,removingmostreferencestooldertechnologies.

• Addeddiskasadismissedtechnology.

• ChangedLTO4toacurrenttechnology.

• AddedT10000B,LTO5,andTS1130asfuturetech-nologies;deletedLTO3,SAIT-1,andSAIT-2.

• Modifiedsothatfuturetechnologiesarenolongerscored.

• Decreasedthenumberofdrivesforscenarios#2and#3.

2 Archive and Records Management—Fiscal Year 2010 Offline Archive Media Trade Study

June 2010

• Overallrefreshofstudy,removingmostreferencestooldertechnologies(T10000,LTO4,DLT).

• ChangedT10000B,LTO5,andTS1130tocurrenttechnologies.

• AddedT10000C,LTO6,andTS1140asfuturetech-nologies.

• Removedmaintenancecostsduetolackofdata.

• Adjustedminimumtransferrateandcapacitytobeconsideredforthestudy.

Introduction

Purpose and Scope

Typically,thepurposeofatradestudyistoanalyzesev-eralcoursesofactionandtoprovidethenecessaryinformationforthesponsortoreachaconclusion.Inothercases,atradestudymayrevalidateanongoingcourseofaction.

Thisdocumentassessestheoptionsforthenextgenera-tionofofflinedigitalarchivestoragetechnologytobeusedforthedigitalarchivesoftheUSGS.Theselectedtechnologymustbecapableofsafelyretainingdatauntilspace,cost,andperformanceconsiderationsdrivethenextmediamigration.Datamustbemigratedbeforeintegritydegrades.

NearlyalloftheUSGSworkingarchiveholdingsnowresideonnearlinerobotictapestorageandarebackedbyanofflinemastercopy.Thenearlinecopyisreferredtoastheworkingcopy.Anongoingneedexistsforofflinestorageforinfrequentlyusedworkingcopies,andmasterandoffsitecopieswheretheworkingcopyisstorednearline.

NotethatLTO4hasbeenthearchivemediaofchoiceatUSGSforthepast2years.LTO5testingwillbegininFY10.ThereisnocompellingreasonfortheUSGStochangetechnologiesawayfromLTOatthistime,andgiventheadvantagesofintergenerationreadcompatibilityinanofflinearchiveenvironment,therewillbeacontinuedinterestin“stayingthecourse”withLTOtechnologyfortheforeseeablefuture.

ThispredispositiontouseLTOtechnologydoesnotnegatetheneedtoperiodicallyrevisitofflinestoragetechnologiestostayinformedofchanges.WhenorifLTOeventuallynolongermeetsUSGSrequirements,thisstudy(infuturerevisions)willhaveshownthewaytotheemergingreplacement.

ThisstudyspecificallydoesnotaddresstheonlineandnearlinetechnologiesusedatUSGS.Theprimarynearlinemass-storagesystemattheEarthResourcesObservationandScience(EROS)CentercontainsanHSMusinganOracleSL8500robotictapelibrary,OracleT10000/T10000Btapedrives,OracleLTO3/LTO4tape

drives,anOraclehostserver,OracleSAMHSMsoftware,andamultivendordiskcache.ThearchitectureofthisHSMwasdeterminedbyatradestudyusingadifferentsetofrequire-mentsthanthisstudy.

ThisstudydeterminesthebestofflinearchivemediatobeusedattheEROSCenterandmeetingUSGScriteria.Thefindingsofthisstudyshouldnotbemisconstruedasananalysisofanyspecifictechnologyforotherpurposessuchasenterpriseorroboticnearlinestorage.Changingthecriteriaweightingfactorswouldproducedifferentfindingstailoredtootherspecificcircumstances.

Background

TheUSGSEROSCenter,inSiouxFalls,SouthDakota,hasarchivedofflinedatasetsusingseveraltechnologies(table1).

In2003,theUSGSmigratedmorethan50,0003480and3490tapestonearlinestorageandto110LTO2tapes.Thismigrationwasperformedoveraperiodof5.5months,slowedbythehandlingofthelargenumberof3480/3490tapes.HighDensityTape(HDT),3480/3490,andDigitalCassetteTape(DCT)wereproventoberobustandhigh-performancefortheirtime.Astechnologyadvances,asdatasetsgrow,asmediaages,andasUSGSDigitalLibraryspacefills,theUSGSmustmigratedatatonewer,morephysicallycompact,andhigherperformingstoragetechnologies.

Data Integrity

Becausetheforemostgoalofanarchiveisdatapres-ervation,dataintegritymustbetheprimarycriteriafortheselectionofthedrivetechnology.Severalelementscontributetodataintegrity:

Table 1. Recent and current archive technologies used at the U.S. Geological Survey.

[Currentinbold.USGS,U.S.GeologicalSurvey;HDT,HighDensityTape;GB,gigabyte;MB/sec,megabytepersecond;MB,megabyte;DLT;DigitalLinearTape;DCT,DigitalCassetteTape;HP,Hewlett-Packard;LTO,LinearTape-Open]

Tape drive technologyYears used at

USGSCapacity Transfer rate Type

HDT 1978–2008 3.4GB 10.6MB/sec Analog3480 1990–2003 200MB 2.0MB/sec Digital3490 1995–2003 900MB 2.7MB/sec DigitalDLT7000 1996–2006 35GB 5.0MB/sec DigitalDCT(AmpexDCRsI) 1992–2007 45GB 12.0MB/sec AnalogSuperDLT220 1998–2008 110GB 10.0MB/sec DigitalOracle 9940B 2002–present 200 GB 30.0 MB/sec DigitalHP LTO Ultrium 2 2003–present 200 GB 40.0 MB/sec DigitalHP LTO Ultrium 3 2005–present 400 GB 80.0 MB/sec DigitalHP LTO Ultrium 4 2007–present 800 GB 120.0 MB/sec Digital

Introduction 3

• Thenumberofarchivalcopies:USGSarchivesmusthaveworkingandmastercopies,andanoffsitecopyisdesirable.Themasterandworkingcopiesneednotbeonsimilarmedia.

• Drivereliability:Aslightlylessreliabledrivetechnol-ogycanbeused,butonlywithasufficientnumberofcopiesinthearchive.

• Thestoragelocationandenvironment:Storagelocationandenvironmentareaconstantforallthetechnologiesassessedbecauseallmediaarestoredinasecureandclimate-controlledenvironment.

• Thecompositionofthemedia:Somemediacomposi-tionslastsignificantlylongerthanothers,butallthetechnologiesinthisstudyusesimilarlong-lastingmediacompositions.

• Tapehandlingwithinthedrive:Thischaracteristicdefineshowatapeishandledbythedrive—whethercontactismadewiththerecordingsurface,howmanyserpentinepassesarerequiredtoreadorwriteanentiretape,andthecomplexityofthetapepath.

• Errorhandling:DrivestypicallyminimizedatalossthroughCyclicRedundancyCheck(CRC)orotherdatarecoverymethods,andallowdatatobereadafterskippingpastanerror.Thougherrordetectiononwriteisrequired,additionalattentiontodatarecoveryonreadisahigherprioritybecausemediadegradationwilleventuallyleadtoreaderrors.

• Primarymarket:Thiscriteriondescribesthetargetmarketofadriveandthecharacteristicsofdrivesinthatmarket(table2).

• Adrivetargetedtothebackupmarketisdesignedforwritemany/readrarelyanddependsmoreonwriteerrordetectionbecausethedataarestillavailableandcanbeeasilyrewritten.Backupdrivesaretypicallybuiltforspeed,capacity,andlowcost.

• Adrivetargetedtotheenterprisemarketisdesignedforwritemany/readmanyuseinaroboticlibraryorauto-stacker,andequalemphasisisplacedondetectingerrorsonreadandwrite.Enterprisedrivesaretypicallybuiltforreliabilityandspeed,withcapacityasecondaryfactor.Costisanotamajorconsideration.

• Adrivetargetedtothearchivalmarketwouldbedesignedforwriteonce/readrarely,andequalemphasiswouldbeplacedondetectingerrorsonreadandwrite;however,nodrivesarecurrentlydesignedormarketedprimarilyforarchiving.Mostvendorswouldarguethattheirproductsarearchivedevices,butifforcedtochoosetheir

primarymarketnovendorwouldchoosethelim-itedarchivemarketoverthelucrativebackuporenterprisemarkets.

Thereliabilityofalong-termarchivetechnologyrelatesprimarilytothelong-termviabilityoftherecordedmedia.Reliabilityintechnologyisdifficulttodetermineexceptinretrospectbecauseatechnologyneedstobeimplementedearlyenoughinthelifecyclethatdrivescanbekeptworkingduringthelifetimeofagivenmedia(orreplacedwithnewerbackward-compatiblemodels).Thisstudybasesthereliabil-ityassessmentonpastexperiencewiththevendorandtheirproducts,onspecifications,ontheexperiencesofothers,orexperiencegainedfrombenchmarking.

Experiencewith3480,3490,9840,9940,andT10000hasshownOracle/Sun/StorageTec(STK)productstobereliable,buttheOracle/SunD3helicalscandrivewasproblematicandwasdiscontinuedquickly.OnseveraloccasionstapesthathadunrecoverableerrorsweresenttoOracleforrecovery.Sometapeswererecovered,butsomewereunrecoverablebecauseofcartridgecontamination.Tapedrivefailurestypicallyhappenwithouttapedamageandarereplacedwithoutcausingdataloss.

Selection Criteria

Thefollowingcriteriawereusedindeterminingwhichtechnologiesshouldbeconsidered.

1. Thetechnologymustbecurrentlyavailableandthemostrecentdrivemanufacturedtobeconsideredinthefinalanalysis.Drivesthatareanticipated/announcedbutnotavailablearementionedbutnotrankedinthefinalanalysis.

2. Thetechnologymusthaveatleast1terabyte(TB)[1,000gigabyte(GB)]capacityofuncom-presseddata.

3. Thetechnologymusthaveanuncompressedwritetransferrateofatleast120megabytespersecond(MB/sec).

Table 2. Tape drive markets and characteristics.

Primary market

Reliability Usage Driving design factors

Backup Moderate Writemany,readrarely

Lowcost,highcapacity,highspeed.

Enterprise High Writemany,readmany

Asmuchas100percentdutycyclefordrivesandmediausedwithrobotics.

Archive High Writeonce,readrarely

Long-termreliability.


4. Thetechnologymustusemediathatcanremainreadableforatleast10yearsinacontrolledenvironment.Thelifetimeof10yearswasselectedbecause10yearsisthelongestthatamediatechnologywouldconceivablybeusedbeforespaceandtransferrateconcernswoulddictateamovetoanewtechnology.

5. Thetechnologymustnotbehamperedbyapoorreliabilityorperformancehistory.Forexample,helicalscantechnologiessuchas4millimeter(mm),8mm,DAT,andD3haveprovenunreli-ableinthepast.

Thefollowingcurrentlyavailabledrivetechnologieswereselectedforconsideration.

1. OracleT10000B;

2. Hewlett-Packard(HP)LTO5(LinearTapeOpen)—representativeofmodelsbyIBM,Quantum,andTandberg;and

3. IBMTS1130.Thefollowingfuturedrivestechnologiesarementioned

butnotconsidered:1. OracleT10000C;

2. HPLTO6;and

3. IBMTS1140.

Dismissed Technologies

Thefollowingtechnologiesweredismissedfromanalysisorconsideration.

Magnetic DiskDiskpricescontinuetodrop,whilereliability,perfor-

mance,andcapacityincrease.Cost,managementoverhead,cooling,andpowerareconsiderationsinusingdisktoarchivelargedatasets.Inthepastseveralyearsithasbecomefeasibletostoretheworkingcopyofsomedatasets,orpartsofdatas-ets,ondiskaslongasarchivecopiesareretained,typicallyontape.Althoughtapecouldstayviableupto10years,themorecostlydiskistypicallyreplacedevery4or5yearstomaintainsupportability,reliability,andperformance.Servingfrequentlyusedworkingcopiesondiskprovidessignificantperformancebenefits,althoughanofflinemastercopymustberetained.

Solid State Disk (SSD)

Similartomagneticdisk,SSDpricescontinuetodrop,whilereliability,performance,andcapacityincrease.ItisexpectedthatSSD,overtime,willreplacemagneticdiskforonlinestorage.SSDdoesoffersomebenefitsregardingarchive

storage—itisexpectedtotoleratelongshelfstoragebetterthanmagneticdisk,whichsuffersfromcoatingdeterioration.EventhoughSSDcouldbecomeanoptionforfutureofflinearchivestorage,itistooexpensivetocompeteatthistime.

CD-ROM, DLT 8000, QIC, Mammoth, and Erasable Optical (EO)

Thiscategoryincludestechnologiesthatarelowcapac-ity,lowperformance,oraged.Alloftheseproductshavebeenavailableforsometimebutcanimmediatelybedismissedonthebasisofobviouslimitationsinperformance,capacity,orreliabil-ity.Theseproductsarenotagoodfitforlargedigitalarchives.

Oracle 9840TheOracle9840isafast-accesstechnologyusedalmost

exclusivelyinconjunctionwithOracleroboticlibraries.Althoughitisanenterprise-classdrive,ithaslowcapacity,lowtransferrate,andhighcost.Theadvantageofthisdriveisthefastaccess:thedualreeldesigndoesnotrequirealengthyloadingsequence,anditispositionedattapemidpointforfasteraccess.Althoughthistechnologyisusefulwherefastnearlineaccessisrequired,thetechnologyoffersminimalbenefitintheofflinearchivemediaarena.

Quantum DLTInpastrevisionsofthisstudy,Quantumpresentedavia-

blechallengetoLTOintheformoftheDLTline.DLThaslostsubstantialmarketsharetothepointthatfurtherdevelopmentofthelinehasbeenofficiallydiscontinued.Althoughdrivesarestillavailable,lackoffurtherdevelopmenthasensuredthatDLTisnolongercompetitivewithLTOandthespecifica-tionsdonotmeettheminimumforthisstudy.QuantumnowproducesLTOdrives.

Tandberg/Exabyte VXA320, Sony SAIT-1/SAIT-2Tandberg/Exabytehasevolvedtheirearlyhelicalscan

technologyintotheVXA320withanativecapacityof160GBandanativetransferrateof24MB/sec.Thistechnologyisbasedonconsumer-gradecartridgeanddrivetechnologies.Althoughmediacostsarelow,transferratesarealsolowandtheUSGSexperiencewithconsumer-gradestoragetechnolo-gieshasshownthatthesetechnologiescannotwithstandtherigorsofalong-termarchive.

Tapedrivessuchasthe8mm/Exabyte,whichbecamepopularinthe1990s,werebasedonconsumer-gradehelicalscantechnologyandwerenotablyslowandunreliable.Longstart/stop/repositioningtimesdictatedthatifdatawerenotkeptstreaming,theeffectivetransferratedroppeddrastically.Thenecessarilycomplexdrivepathledtoproblems:8mmdrivesmangledtapes,andaconfusingarrayoffirmwarever-sionsoftenyieldedunpredictablebehaviorandhangs.The

Technical Assessment 5

transitionfromamarketonceruledby4mm/8mmhelicalscandrivestooneruledbyLTO/DLThappenedquickly,andthesmallcurrentmarketshareofhelicalscantechnologiesmayindicatethatthemarketplacestillremembersthedif-ficultiesofearlierhelicalscandrives.Themarketmayneverreconsiderwhethertheearlierproblemsareovercomeunlessnewterminologyreplaces“helicalscan.”

TheSonySAIT-1andSAIT-2seemedpromisingwhenfirstannouncedbutwerelatetomarket,haveslowtransferrates,andnevergainedsufficientmarketsaturationtolowermediacosts.TheSAIT-2isreportedlyonlyavailableinaSonyroboticlibrary,whichistargetedtovideoautomationinthetelevisionindustry.

DVD, HD-DVD, Blu-RayDigitalVideoDisc(DVD)andrelatedtechnologiesseem

promisingfromthestandpointofexpectedlongevityofthemedia;however,studieshaveshownthatopticalmediacandegradeandbecomeunusableinaslittleas5years.Lowcapacitypermedia,lowtransferrates,lackofmediaprotec-tion(noshell),nosinglestandard,andhighmediacostsadduptoaproductthatsimplywouldnotworkforhighvolumearchivaluse.

HighDefinitionDigitalVersatileDisc(HD-DVD)waswithdrawnfromthemarketplaceafterfailingtocompetewithBlu-Ray.Blu-Raywouldcertainlyhavesomeapplicationindistributionandshort-termstorageoflargeamountsofdata,butlikeCompactDisc(CD)andDVD,Blu-Raysuffersfromhighmediacostsandlowtransferrates,andgivenopticalmediahis-tory,theshelflongevitymustbeprovenbeforebeingtrustedinanarchiveenvironment.

Newer Optical Technologies

Severalhigh-capacityopticaldisktechnologieshavebeeninthedevelopmentphaseforthepastfewyears.Ofthetechnol-ogyproposalsthathaveappearedintradejour-nalsandatconferences,noneareavailable.

Onehigh-techexampleoffuturetech-nologiesisholographicstorage.Productshavebeenrepeatedly

announced,buthaveyettoship.HolographicVersatileDisc(HVD)specificationsindicateaplannedcapacityof3.9TBperdiskandatransferrateof125MB/sec.Rivalsclaimasmuchas100TBperdiskwillbepossible.

Technical Assessment

Analysis

Thistechnicalassessmentincludesdrivesselectedforfinalevaluation(T10000B,LTO5,TS1130)anddrivesanticipatedtobereleasedinthenearfuture(T10000C,LTO6,TS1140)(table3).LTOdrivesareavailablefrommultiplevendors(Tandberg,Quantum,IBM,HP),withHPselectedtorepresentLTOtechnologyinthisstudy.Thefollowingtapetechnologieswillbeevaluated,butonlythedrivesshowninboldwillbeincludedinthefinalevaluation.

• Oracle T10000B

• OracleT10000C

• HP LTO5

• HPLTO6

• IBM TS1130

• IBMTS1140

Table 3. Technology comparison.

[Yellowhighlightedtextindicatesunverifiedinformation.HP,Hewlett-Packard;LTO,LinearTape-OpenTB,terabyte;MB/sec,megabytepersecond;TBD,tobedetermined;m/sec,meterspersecond;HW,hardware;MB,megabyte;GB,giga-byte;est,estimated]

Specification T10000B T10000C HP LTO5 HP LTO6 TS1130 TS1140Uncompressedcapacity 1.0TB 2.0TB 1.5TB 3.2TB 1.0TB 2.0TBUncompressedxferrate 120MB/sec 200+MB/sec 140MB/sec 210MB/sec 160MB/sec 240MB/secRecordingtechnology Serpentine Serpentine Serpentine Serpentine Serpentine SerpentineTracks 1,152 TBD 1,280 TBD 1152 TBDChannels 32 32 16 TBD 16 TBDPasses 36 TBD 80 TBD 72 TBDTapevelocity 3.74m/sec TBD TBD TBD 8.6m/sec TBDType Enterprise Enterprise Backup Backup Enterprise EnterpriseEncryptionsupport HWoption HWoption HWbuilt-in HWbuilt-in HWbuilt-in HWbuilt-inBuffersize 256MB 256MB 256MB TBD 1GB 1GBAdaptivespeeds 2speeds 2speeds 47–140MB/s Dynamic 6speeds MultiplePrice $24,000 $24,000est $3,200 $3,200est $29,000 $29,000estShelvescompatible? Yes Yes Yes Yes Yes YesPreviousgenerationsread 1 TBD 2 2 1 TBDPreviousgenerationswritten 0 TBD 1 1 1 TBDBitErrorRate(BER) 1x10-19 1x10-19 1x10-17 1x10-17 1x10-17 1x10-17

Drivemanufacturers 1 1 4+ 4+ 1 1Availability 2008 Late2010 2010 2012 2008 2011


Oracle T10000B

TheT10000BistheOracleflagshiphigh-capacityenter-prisedrivetypicallyusedinconjunctionwithOracleroboticlibraries,suchastheSL8500.TheEROSCenterhaseightT10000BdrivesforuseintheSL8500,inadditiontosixfirst-generationT10000drives(fig.1).

Advantages

• TheT10000Bisanevolutionofthe9940,whichtheUSGShasdeterminedtobeextremelyreliable.

• Nativecapacityis1TBandnativetransferrateis120MB/sec.TheT10000Balsocanstreamat50MB/sec,whichisimportantbecausesomedisksmaynotbeabletokeepupat120MB/sec.

• TheT10000Buses32channelsperpass(comparedto16oncompetingdrives),whichreducesserpentinepasses.With1,152tracks,only36passesarerequiredtoreadorwritetheentiretape.

• TheT10000Bistargetedtotheenterprisestoragemar-ketwheredataviability,speed,andcapacityaremoreimportantthancost.

• TheT10000Bwasdesignedasarobuststoragemedia,withthetapecartridgeanddrivebuilttowithstandconstantorfrequentuseinaroboticenvironment.ThedrivesarecompatiblewiththeSL8500andexcelinaroboticenvironmentbecauseoftheirdurability.

• T10000Bdrivesprovidedrivestatisticsforservoerrors,bytesread/written,I/Oretries,andpermanenterrors.

• T10000BusesthesamemediaastheT10000,allow-ingmediare-use.TapeswritteninT10000BformatcannotbereadbytheT10000drives.

• TheT10000Bhasa256MBbuffer,whichpreventsocca-sionaldatastarvationfromreducingthetransferrate.

• TheBitErrorRate(BER)isanindustrybestat1x10-19.

• Ahardwareencryptionoptionmoduleisavailable.

Disadvantages

• TheonlycartridgesavailableareproducedforOraclebyImationandFuji.

• TheT10000Bdrivesare7timesthepriceoftheLTO5butcheaperthantheTS1130.

• BasedonsalesoftheT10000theT10000BsalesareanticipatedtobeprimarilyforuseinOraclerobotics.Forthisreason,theT10000BisanticipatedtohaveamarketsharethatwillremainlowcomparedtoLTO,ensuringthatmediacostswillremainhigh.

• TheT10000BdriveisonlyavailablefromOracle.Thisavailabilitykeepsthepricehighbutdoeseliminateconcernsofincompatibility.

SummaryTheT10000Bisahigh-capacity,high-transferrate,enter-

prise-classdriveforuseinOracleroboticlibraries.ThecostofmediaanddrivesfarexceedsthecostofLTO,butmediareuseforfuturegenerationswouldeffectivelyreducemediacosts.Therobusttechnologywouldbeaprimechoiceforofflinearchivesifonlyonecopyofadatasetcouldbekept.Whentwoormorecopiesofadatasetexist,andoneisalreadyonanenterprisetechnologysuchasT10000B,useofanenterprisesolutionforthesecondcopyisnotwarranted.

Oracle/Sun

Figure 1. Screen capture showing the Oracle roadmap (uncompressed)


Oracle T10000C

TheT10000CisthethirdgenerationoftheT10000line.TheT10000Cwasoriginallyanticipatedtoshipinspring2010,butisnowexpectedtoshipinlate2010orearly2011.

Advantages

• TheT10000CisanevolutionoftheT10000/T10000B,whichtheUSGShasdeterminedtobeextremelyreliable.

• Nativecapacityisanticipatedtobeatleast2TBandnativetransferrateofatleast200+MB/sec.TheT10000Cisexpectedtostreamatlowerrates,whichisimportantbecausesomedisksmaynotbeabletokeepupat200+MB/sec.

• TheT10000Cisexpectedtouseatleast32channelsperpass(comparedto16oncompetingdrives),whichreducesserpentinepasses.

• TheT10000Cistargetedtotheenterprisestoragemar-ketwheredataviability,speed,andcapacityaremoreimportantthancost.

• ThemediafortheT10000CisexpectedtodifferfromthemediafortheT10000/T10000B,buttheT10000CmaybeabletoreadmediawrittenonT10000/T10000B.LiketheT10000media,theT10000Cmediawilllikelybedesignedasarobuststoragemedia,withthetapecartridgeanddrivebuilttowithstandconstantorfrequentuseinaroboticenvironment.ThedrivesareexpectedtobecompatiblewiththeSL8500.

• T10000Cdrivesshouldprovidedrivestatisticsforservoerrors,bytesread/written,I/Oretries,andpermanenterrors.

• Somefuturefollow-ondrivesareexpectedtousethesamemedia,allowingmediareuse.

• TheT10000Cisexpectedtohaveatleasta256MBbuffer,whichpreventsoccasionaldatastarvationfromreducingthetransferrate.

• TheBERisexpectedtobeanindustrybestat1x10-19.

• Ahardwareencryptionoptionmoduleisanticipated.

Disadvantages

• CartridgesmaybesuppliedonlybyOracle.

• TheT10000Cdrivesareexpectedtobe7timesthepriceoftheLTOandcheaperthantheTS1130.

• BasedonsalesoftheT10000,theT10000Cisantici-patedtobeprimarilyforuseinOraclerobotics.For

thisreason,themarketshareisanticipatedtoremainlowcomparedtoLTO.

• TheT10000CdriveisexpectedtobeavailableonlyfromOracle.Thisavailabilitykeepsthepricehighbutdoeseliminateconcernsofincompatibility.

SummaryTheT10000CshouldreplacetheT10000/T10000Bdrives

astheflagshiphigh-capacityenterprisedrivetypicallyusedinconjunctionwithOracleroboticlibrariesbecausetheT10000Cshouldbepricedcomparably.TheT10000Cisnotyetavail-ableand,therefore,wasnotassessedinthefinalevaluation.

HP LTO5

TheLTO5isthemostrecentavailablegenerationoftheLTOtapefamily(fig.2)andwillbetestedattheEROSCenterinlate2010.

Advantages

• LTOhasenjoyedphenomenalgrowthfromthedayofreleasein2000;asof2006,LTOheldan82percentmarketshare.Sincethen,furtherdevelopmentoftheleadingcompetingproducts(DLTandSAIT)hasbeendiscontinued.Since2006,LTOdroveDLTandSAIT-2fromthemarket.

• Nativecapacityis1.5TBandnativetransferrateis140MB/sec.

• TheHPLTO5drivecanadaptthetransferratetomatchthestreamingspeedofthesource.

• LTO5isbackwardreadcompatiblewithLTO3andLTO4,andbackwardwritecompatiblewithLTO4(atthelowerLTO4density).

• LTOwasdevelopedbyaconsortiumofHP,IBM,andQuantum(acquiredfromSeagate/Certance)andislicensedtoothers,includingmediamanufacturers.Thiswideacceptancehasintroducedcompetition,whichhasinturncontrolledcosts.

• TheLTO5hasa256MBbufferthatpreventsocca-sionaldatastarvationfromreducingthetransferrate.

• Hardwareencryptionisavailable.

Disadvantages

• LTOistargetedtothebackupmarketwherespeed,capacity,andcostaremoreimportantthanlong-term


integrityofthedata.Becausebackuptapesarewritemany/readrarely,errorswouldlikelyshowupinawritepasswheretheerrorscanbeworkedaround(rewrites)orthemediadiscarded.

• Repeatedend-to-enduseofatapewouldbeaconcernbecauseoneend-to-endread/writeincurs80passes(1,280tracksdividedby16channels).Thisrepeateduseshouldnotbeaconcernforarchiveoperationsbecauseusageislimited.

• EachgenerationofLTOrequiresnewmediatoattaintheratedcapacity,ensuringthatmediacostswillbesubstantiallyhigheruntilmarketsaturationdrivesthepricedown.Thepriceshouldnotbeaconcernforarchiveoperations,becauserequiredmedialifeistypicallysupportedbydrivebackwardcompat-ibility.

• LTOwasdesignedasamoderateusagestoragemedia,withthetapecartridgeanddrivenotbuilttowithstandconstantenterprise/roboticuse.

• LTOwasco-developedbyIBM,HP,andQuantum(acquiredfromSeagate/Certance).Thiskindofpart-nershipmakesitpossibleforeachvendortointerpretthespecificationsdifferentlyandtodesigndrivesthatmayhaveincompatibilities,thoughcompatibil-itytestsareperformed.EROSobservedtwoLTO1incompatibilityproblemsbetweenHPandIBM:tapeswrittentoend-of-tape(EOT)ontheIBMcannotbereadontheHP,andtapeswrittenontheHPreadatlessthanhalfspeedontheIBM.Inter-brandincom-patibilitiescanbeavoidedbyusingasinglebrandofdrive.

SummaryTestingofLTO5technologyatEROSwillbegininlate

2010.LTOhasbeenreliableatUSGS,withonlyasmallnum-beroffailurescommensuratewiththedesignspecificationsforamid-rangetapetechnology.

HP LTO6

TheLTO6isthenextanticipatedgenerationoftheLTOtapefamily,withreleaseanticipatedin2012basedonatypicalLTOreleasecycleof2years.

Advantages

• LTOhashadphenomenalgrowthfromthedayofreleasein2000;asof2006,LTOheldan82percentmarketshare.

• Nativecapacityisexpectedtobe3.2TBandnativetransferrateisexpectedtobe210MB/sec.

• TheHPLTO6driveisanticipatedtouseanadaptivetransferratetomatchthestreamingspeedofthesource.

• LTO6shouldbebackwardreadcompatiblewithLTO4andLTO5,andbackwardwritecompatiblewithLTO5(atthelowerLTO5capacity).

• LTOwasdevelopedbyaconsortiumofHP,IBM,andQuantum(acquiredfromSeagateCertance)andislicensedtoothers,includingmediamanufacturers.Thiswideacceptancehasintroducedcompetition,whichhasinturncontrolledcosts.

• Hardwareencryptionisanticipated.

LTO Consortium

Figure 2. Screen capture showing the LTO roadmap (with 2:1 compression)


Disadvantages

• LTOistargetedtothebackupmarketwherespeed,capacity,andcostaremoreimportantthanlong-termintegrityofthedata.Becausebackuptapesarewritemany/readrarely,errorswouldlikelyshowupinawritepasswheretheerrorscanbeworkedaround(rewrites)orthemediadiscarded.

• Repeatedend-to-enduseofatapewouldnormallybeaconcernbecauseoneend-to-endread/writeisexpectedtoincur80ormorepasses.Thisrepeateduseshouldnotbeaconcernforarchiveoperationsbecauseusageislight.

• EachgenerationofLTOrequiresnewmediainordertoattaintheratedcapacity,ensuringthatmediacostswillbesubstantiallyhigheruntilmarketsaturationdrivesthepricedown.Thepriceshouldnotbeaconcernforarchiveoperationsbecauserequiredmedialifeistypi-callysupportedbydrivebackwardcompatibility.

• LTOwasdesignedasamoderateusagestoragemedia,withthetapecartridgeanddrivenotbuilttowithstandconstantuse.

• LTOwasco-developedbyIBM,HP,andQuantum(acquiredfromSeagate/Certance).Thiskindofpart-nershipmakesitpossibleforeachvendortointerpretthespecificationsdifferentlyandtodesigndrivesthatmayhaveincompatibilities,thoughcompatibilitytestsareperformed.EROSobservedtwoLTO1incompat-ibilityproblemsbetweenHPandIBM:tapeswrittentoEOTontheIBMcannotbereadontheHP,andtapeswrittenontheHPreadatlessthanhalfspeedontheIBM.EROSresolvedthisissuebyonlydeployingHPdrivesforproductionuse.

SummaryLTO6isexpectedtobe

announcedin2011andmadeavail-ablein2012.LTO6isnotyetavail-ableandwasnotassessedinthefinalevaluation.

IBM TS1130

TheTS1130isanenterprise-classtapedrive,usedprimarilyinroboticlibrariesandautoloaders.TheTS1130isafollow-ondrivetotheTS1120(fig.3).

Advantages

• Lineageincludesthereliable3480,3490,3590,3592,andTS1120.

• Supportsa4gigabitpersecond(Gbit/sec)FiberChannelinterface.

• Nativecapacityis1TBandnativetransferrateis160MB/sec.

• TheTS1130isarobuststoragetechnology,withthetapecartridgeanddrivebuilttowithstandconstantorfrequentuseinaroboticenvironment.

• TheTS1130usesthesamemediaastheTS1120and3592,plusanewhighercapacitycartridge.

• Ahardwareencryptionfeatureisincludedinthedrive.

Disadvantages

• DesignedprimarilyforuseinIBMroboticlibraries.

SummaryTheTS1130doesnotcomparefavorablyincosttoLTO,

andenterprise-classrobustnessisnotrequiredwhenthework-ingcopyofadatasetisalreadyonenterprise-classtechnologyintheUSGSroboticlibrary.IBMrecentlyreporteddevelop-mentofarecordingmethodthatwillyieldacapacityof35TBpercartridge,butIBMdidnotrevealatimeline.

International Business Machines

Figure 3. Screen capture showing the IBM roadmap (uncompressed)


IBM TS1140

TheTS1140isanticipatedtobethenextgenerationofthe3592tapefamily,withreleaseexpectedin2011.NotethattheTS1140namehasnotbeenconfirmed,butfollowslogically.

Advantages

• Lineageincludesthereliable3480,3490,3590,3592,TS1120,andTS1130.

• Shouldsupporta4or8Gbit/secFiberChannelinter-face.

• Nativecapacityisexpectedtobe2TBandnativetransferratemayexceed240MB/sec.

• TheTS1140willbearobuststoragetechnology,withthetapecartridgeanddrivebuilttowithstandconstantorfrequentuseinaroboticenvironment.

• TheTS1140mayusethesamemediaastheTS1130.

• Ahardwareencryptionfeatureshouldbeincludedinthedrive.

Disadvantages

• DesignedprimarilyforuseinIBMroboticlibraries.

SummaryTheTS1140wouldnotcomparefavorablyincosttoLTO,

andenterprise-classrobustnessisnotrequiredwhenthework-ingcopyofadatasetisalreadyonenterprise-classtechnologyintheUSGSroboticlibrary.TS1140isnotyetavailableandwasnotassessedinthefinalevaluation.

Tables

Design Criteria

Thedesigncriteriaandtargetmarketofadriveareinter-related(table4).LTO5istargetedtothebackupmarket,asdemonstratedbyLTOmarketing.TheT10000BandTS1130aretargetedtotheenterprise(datacenter)market.

Adrivetargetedtothebackupmarketisdesignedforwritemany/readrarelyanddependsonwriteerrordetectionbecausethedataarestillavailableandcanbeeasilyrewritten.Backupdrivesaretypicallybuiltforspeed,capacity,andlowcost.

Adrivetargetedtotheenterprisemarketisdesignedforwritemany/readmanyuseinaroboticlibraryorauto-stacker,andequalemphasisisplacedondetectingerrorsonreadandwrite.Enterprisedrivesaretypicallybuiltforreliabilityandspeed,withcapacityasecondaryfactor.Costisnotamajorconsiderationtoenterpriseuserswillingtopayforquality.

Adrivetargetedtothearchivalmarketwouldbedesignedforwriteonce/readrarely,andmoreemphasiswouldbeplacedondetectingandcorrectingerrorsonread;however,therearecurrentlynodrivesdesignedormarketedprimarilyforarchiveuse.

Theformulausedtorankdesigncriteriawas:

((100-serpentinepasses)/10)+(absolutevalueoferrorrateexponent/2)+

(construction3=moderateusage,5=highusage)+(headcontact3=contact,5=mincontact)/2.59(toadjustthehighestrankto10)

Transfer Rate

Transferrateisimportantbecauseitestablisheshowquicklythemigrationofanarchivedatasetmaybecompleted

Table 4. Design criteria and target market.

[Uncorrectederrorratesforsomedrivesarenotavailablebutarepresumedtobeeitherthesameastheirpredecessororatleast1x10-17.Yellowhighlightedtextindicatesunverifiedinformation.MP,metalparticle;TBD,tobedetermined;HP,Hewlett-Packard;LTO,LinearTape-Open;IBM,InternationalBusinessMachines]

TechnologySerpentine

tracks/ passes

Target market

Tape composition

Uncorrected error rate

Cartridge construction

ratingHead contact Ranking

OracleT10000B 1,152/36 Enterprise AdvancedMP 1x10-19 Highusage Minimumcontact 10.0OracleT10000C TBD Enterprise AdvancedMP 1x10-19 Highusage Minimumcontact 10.0HPLTO5 1,280/80 Backup ThinfilmMP 1x10-17 Moderateusage Contact 6.4HPLTO6 TBD Backup ThinfilmMP 1x10-17 Moderateusage Contact 6.4IBMTS1130 1,152/72 Enterprise AdvancedMP 1x10-17 Highusage Contact 8.2IBMTS1140 TBD Enterprise AdvancedMP 1x10-17 Highusage Contact 8.2

Tables 11

andhowfastaproductionsystemmaygenerateproductsfromthearchivemedia.Theminimumtransferraterequirementis120MB/sec,with140MB/secdesired.MuchofthedataarchivedattheUSGSarerasterimagerythattypicallylackrepeatablepatternsthatwouldcompresswell;therefore,alltransferratescitedarenative(uncompressed).

Wheremeasuredtransferrateswerenotavailable,approximateratesaredeterminedbasedontheaccuracyofspecifiedtransferratesofpreviousgenerations.Thesourceofthetestresultsalsoappliestocapacitiesintable5.

Therankingwasdeterminedbyaddingtheactual/approx-imatereadandwriteratesforeachdrive,settingtherankingforthefastestdriveto10,thenrankingtheothersagainsttheleader.Forexample,adrivehavinghalfofthetotalread/writetransferrateoftheleaderwouldberanked5.

Capacity

Asecondaryrequirementistoconserverackorpal-letstoragespaceandreducetapehandlingbyincreasingpermediacapacity.ThecurrentarchivemediaofchoiceattheUSGSisLTO4at757GBofusablecapacitypertape.Thenewminimumcapacityrequirementis1TB,with1.5TBor

moredesired.Allthereviewedtechnologiesmeetthe1TBrequirementbasedontheadvertisedcapacity.Becausemuchofthedataarchivedarenotcompressible,allcapacitiesarenative(uncompressed).Wheremeasuredcapacitieswerenotavailable,approximatecapacitiesaredeterminedbasedontheaccuracyofspecifiedcapacitiesofpreviousgenerations.

Thecapacitieslistedintable6presumethatagigabyte=1,073,741,824bytes.Theratingsweredeterminedbycomput-ingeachasthepercentageofthehighestcapacitydriveonascaleof1to10,withthehighestcapacityasa10.Thesourceofthecapacityratingsarenotedintable6.Notethatcapacityyieldvariesbymediavendor.

Cost Analysis

Table7showstherelativedriveandmediacosts,drivewarranty,andthecostperterabyteformedia.Rankingswereestablishedbysettingthecheapest(driveandmedia)to10thenratingeachoftheothersagainstthelowestcost.Mediacostsperterabytearebasedonadvertisedcapacity.Costsdonotincludesysteminterfacesorcables.PricesarebasedonthelowestpricefoundontheWeborongovernmentpricelists.

Table 5. Transfer rates.

[Yellowhighlightedtextindicatesunverifiedinformation.EROS,EarthResourcesObservationandScience;HP,Hewlett-Packard;LTO,LinearTape-Open;IBM,InternationalBusinessMachines;MB/sec,megabytepersecond]

Tape drive technology

Advertised/ proposed native

rate

Source of test results

Actual/approximate native write transfer rate

Advertised capacity

(in percent)

Actual/approximate native read transfer rate

Advertised capacity

(in percent)Ranking

OracleT10000B 120MB/sec EROStesting 109.00MB/sec 91 120.00MB/sec 100 7.4OracleT10000C 180MB/sec Approximate 163.80MB/sec 91 180.00MB/sec 100 7.4HPLTO5 140MB/sec Approximate 126.70MB/sec 90.5 126.56MB/sec 90.4 8.2HPLTO6 210MB/sec Approximate 190.05MB/sec 90.5 189.84MB/sec 90.4 8.2IBMTS1130 160MB/sec Vendor 153.92MB/sec 96.2 153.92MB/sec 96.2 10.0IBMTS1140 240MB/sec Approximate 230.88MB/sec 96.2 230.88MB/sec 96.2 10.0

Table 6. Storage capacities.

[Yellowhighlightedtextindicatesunverifiedinformation.TB,terabyte;GB,gigabyte,HP,Hewlett-Packard;LTO,LinearTape-Open;IBM,InternationalBusinessMachines]

Tape drive technologyAdvertised/

proposed native capacity

Measured/approximate native capacity

Advertised capacity

(in percent)Ranking

OracleT10000B 1.0TB 936GB 93.6 6.6OracleT10000C 2.0TB 1,872GB 93.6 6.6HPLTO5 1.5TB 1,420GBapproximate 94.7approximate 10.0HPLTO6 3.2TB 3,030GBapproximate 94.7approximate 10.0IBMTS1130 1.0TB 950GBapproximate 95.0approximate 6.7IBMTS1140 2.0TB 1,900GBapproximate 95.0approximate 6.7


MaintenanceshouldbeaconsiderationbutwasremovedfromthisiterationofthestudybecauseofthetenuousstatusofOraclesupportcostsasofthiswriting,andincompleteinfor-mationonLTOsupport.Maintenancewillbereconsideredinthenextupdate.

3.5 Scenarios

Table8showsthetotaldriveandmediacostforthreescenarios.Thesescenariospresumethateachdatasetorprojectstandsalone,althoughpoolingresourcesformultipledatasetscanmitigatecost.Competitionoftenresultsinaconsider-abledropinmediapriceswithin6monthsafterproductintroduction.

Rankingsarebasedonthe100TBoptionandwereestablishedbysettingthecheapestto10thenratingeachoftheothersagainstthelowestcost.Advertised/proposednativecapacitiesareused.Costsdonotincludemaintenance,systeminterfaces,orcables.

Thoughnotrepresentedinthisstudy,technologyrefreshcostsrelatedtomovingfromonegenerationtothenextmayvarydependingonwhetherthevendorrequiresamediachange.LTOhasalwaysrequirednewmediaforeachgenera-tion,butOracleandIBMtypicallyhaveusedthesamemediaforatleasttwogenerations.

Vendor Analyses

Table9providesananalysisofeachcompanyandthestabilityofeachtechnology.Allareestablishedandstablecompanies;therefore,thisratingshouldnotbeviewedasamarketanalysis.Whenselectinganarchivetechnology,itmakessensetolookatthecompanyandproducthistorieseventhoughratingvendorhistoryischallengingbecauseofmergersandacquisitions.ForT10000B,thetechnologywasbasedonthepredecessor9940;therefore,thetechnologyageincludesthe9940.Thelongevityrankingsweredeterminedbythefol-lowingformula:

(companyage+technologyage)/11.4(toadjustthehighestrankto10)

Determiningcompanyyearsinbusinessiscomplicatedbymergersandacquisitions,suchaswhenSunacquiredSTKandwaslateracquiredbyOracle.TheyearsinbusinessbeganwithSTKbecausethetapetechnologyofferedtodayisbasedonSTKproducts.Thepurposeofthissectionistoassesstech-nologylineageandcompanyhistory,butmergersandacquisi-tionsmaybedistractiveanddetrimentalwhenconsideringlineageandhistory.

Drive Compatibility

Table10showsthelevelofintergenerationdrivecompat-ibilityandthefuturedrivesplanned.Thecolumns“Percentageofpreviousgenerationsread”and“Percentageofprevious

Table 7. Drive and media costs.

[Yellowhighlightedtextindicatesunverifiedinformation.TB,terabyte;est,estimated:HP,Hewlett-Packard;LTO,LinearTape-Open;IBM,InternationalBusinessMachines]

TechnologyDrive

(dollars per each)Drive

warranty

Media (dollars

per each)

Media (dollars per TB)

Ranking drive cost

Ranking media cost

per TBOracleT10000B $24,000 12month $125 $125 1.3 5.6OracleT10000C $24,000est 12month $125est $62est 1.3 5.6HPLTO5 $3,200 36month $105 $70 10.0 10.0HPLTO6 $3,200est 36month $105est $33est 10.0 10.0IBMTS1130 $29,000 12month $178 $178 1.1 3.9IBMTS1140 $29,000est 12month $178est $89est 1.1 3.9

Table 8. Scenario costs (drives, media).

[Yellowhighlightedtextindicatesunverifiedinformation.TB,terabyte;HP,Hewlett-Packard;LTO,LinearTape-Open;IBM,InternationalBusi-nessMachines]

Technology100 TB

2 drives200 TB

3 drives400 TB

4 drives100 TB

ranking

OracleT10000B $60,500 $97,000 $146,000 2.2OracleT10000C $54,250 $84,500 $121,000 2.2HPLTO5 $13,435 $23,565 $40,730 10.0HPLTO6 $9,700 $16,200 $26,000 10.0IBMTS1130 $75,800 $122,600 $187,200 1.8IBMTS1140 $66,900 $104,800 $151,600 1.8

Table 9. Vendor analyses.

[STK,StorageTek;HP,Hewlett-Packard;LTO,LinearTape-Open;IBM,InternationalBusinessMachines]

Company TechnologyYears in business

Technology age, in years

Longevity ranking

Oracle/Sun/STK T10000 41(1969) 10(2000) 4.5HP LTO 71(1939) 10(2000) 7.1IBM 3592(3590) 99(1911) 15(1995) 10.0

Conclusions and Recommendations for USGS Offline Archiving Requirements 13

generationswritten”indicatethepercentageofpreviousgenerationsthatareread/writtenbythegenerationindicated.Drivesthatarethefirstgenerationreceiveascoreof50per-cent,sothefirstgenerationproductwillnotbepenalized.Thecolumn“Futuregenerationsplanned”indicatesthenumberofgenerationsplannedinthecurrentdrivefamily,followingthedriveindicated.Therankingwasdeterminedbythefollowingformula:

(Percentageofpreviousgenerationsread+Percentageofpreviousgenerationswritten+

(Futuregenerationsplannedx20))/21(toadjust

thehighestrankto10)

Ranking Summary

Therankingsummaryprovidesaquickreferencetotherankings(table11).

Conclusions and Recommendations for USGS Offline Archiving Requirements

Weighted Decision Matrix

Table12providesaweightedanalysisofthedrivesconsidered.Thecriteriaemphasizetheimportanceoftraitscontributingtodatapreservation.TheUSGSmadethefinaldecisionregardingwhichcriteriatouseandtherelativeweightingofthecriteria.Thecolumnsingreenarerelativeratingsforeachtechnology.Thecolumnsinyellowarecalcu-latedbymultiplyingtherelativeweightbytherelativerating.Thefollowinglistdescribeseachcriterion:

• Design(reliabilityofmedia):Thiscriteriondescribestheabilityofthemediatoremainreadableovertime.Includedinthiscriterionisthenumberofpassesperfull-tapereadorwrite,cartridgeconstruction,uncor-rectedBER,andamountofheadcontact(table4).

• Capacity:Thiscriteriondescribesthemeasuredorapproximatecapacitypercartridge,whichistypicallylessthantheadvertisedcapacity(table6).

• Mediacost/TB:Thiscriterionisaratingoftherelativecostperterabyteformediausingtheadvertisedcapac-ity(table7).

• Compatibility:Thiscriteriondescribesthelikelihoodthatthedrivetechnologywillcontinuetoevolveandtheextenttowhichfuturedriveswillhavebackwardreadandwritecapability.Thiscriterionwillgiveanindicationoftheabilitytomaintaindrivesthatcanreadanagingarchive(table10).

• Transferrate:Thiscriteriondescribestheaggregatereadandwritetransferrate,whichistypicallylessthantheadvertisedtransferrate(table5).

Table 10. Drive compatibility.

[Yellowhighlightedtextindicatesunverifiedinformation.HP,Hewlett-Pack-ard;LTO,LinearTape-Open;IBM,InternationalBusinessMachines]

Technology

Percentage of previous generations

read

Percentage of previous generations

written

Future generations

plannedRanking

OracleT10000B 100 0 3 7.6OracleT10000C 100 0 2 7.6HPLTO5 50 25 3 6.4HPLTO6 40 20 2 6.4IBMTS1130 100 50 3 10.0IBMTS1140 100 50 2 10.0

Table 11. Ranking summaries.

[Blueindicatesthehighestrankingincategory.HP,Hewlett-Packard;LTO,LinearTape-Open;IBM,Interna-tionalBusinessMachines]

DriveDesign criteria

CapacityMedia

costDrive

compatibilityTransfer

rateDrive cost

Vendor analyses

Scenario cost

T10000B 10.0 6.6 5.6 7.6 7.4 1.3 4.5 2.2HPLTO5 6.4 10.0 10.0 6.4 8.2 10.0 7.1 10.0IBMTS1130 8.2 6.7 3.9 10.0 10.0 1.1 10.0 1.8


• Drivecost:Thiscriterionistheratingofrelativecostofeachdriveatthelowestcurrentlyavailableprice(table7).

• Vendoranalyses:Thiscriterionistheratingoftheviabilityofthevendorandtechnology(table9).

• Scenariocost:Thiscriterionistheratingofthecostofscenario#1,whichcomprisesmediacostanddrivecost.Themeasuredorapproximatecapacityisusedratherthanadvertisedcapacity(table8).

Notethatinthedecisionmatrixspreadsheetlistedintable12,notallcriteriahavebeenselectedforthefinalanalysisofthistradestudy.Theseunusedcriteriawereleftinthespread-sheetsothatusersmayinsertthecriteriaweightsfortheirspecificapplication.

Conclusions and Notes

LTO5achievedthehighesttotalscoreinthestudy;there-fore,nocompellingreasonexiststoabandonLTOtoadoptanewstandardofflinearchivetechnology.

LTO5andTS1130werenotavailabletobetestedforthisstudy;therefore,performanceandcapacityfigureswerebasedonvendororcustomerbenchmarkswhereavailableorondrivespecificationscombinedwithpastperformance(percent-ageoftheclaimedspecificationsthatwereachievableinthepast).

• Whenmultiplecopiesofadatasetaremaintained,trad-ingcostandperformanceforreliabilityisacceptable,particularlywhentheworkingcopyisonanenterprisetechnologysuchasOracleT10000B,asaremostarchivesatUSGS.

• Asanydrivesaturatesthemarket,mediaanddrivecostsdrop.BasedonUSGSexperiencewithenterprisetapetechnologyandobservationofOracleandIBM

pricing,enterprisedrivessuchastheT10000BandTS1130areunlikelytoachievealevelofmarketsatu-rationthatwouldcausesubstantialpricedecreases.

• Withproperhandlingandmultiplecopies,anyofthetechnologiesevaluatedinthisreportcouldbedeployedforarchiveuse.Whenmorethantwocopiesexist,allcouldbeonnon-enterprisetechnology.

Recommendations

1. TheUSGSshouldcontinuewithLTO4astheofflinestoragemediaofchoice,thentestandmovetoLTO5,whenavailable.

2. DatastoredonLTO2andLTO3shouldbemigratedtoLTO5inthenext2years.

3. Toreducerisk,theUSGSshouldcontinuethestrategyofstoringdatasetsonmultipletechnologieswhenonlytwocopiesexist.Forexample,storeaworkingcopyofadatasetonnearlineT10000Bandoffline/offsitecopiesonLTO.Thisstrategypartlymitigatestherisksofoneortheothertechnologyfailingorbeingretiredprematurely.

4. Inadditiontoanearlineandoffsitecopyofadataset,anonsiteofflinecopyshouldbemaintained,provid-ingfastrecoverywithoutriskingtheshippingoftheoffsiteLTOcopy.

5. TheUSGSshouldadoptapolicyofperiodicallytest-ingarchivetapesforreadability.ThistestingshouldnotbeextensiveenoughtoincurunduewearonthemediaorfrustratetheNationalArchivesandRecordsAdministration(NARA),butshouldbefrequentenoughtoprovideanopportunitytodetectdeterio-ratingmedia.

Table 12. Weighted decision matrix.

[TB,perterabyte]

Selection criteria WeightOracle

T10000BHP

LTO5IBM

TS1130Oracle

T10000BHP

LTO5IBM

TS1130

Designcriteria 0 10.0 6.4 8.2 0.0 0.0 0.0Capacity 20 6.6 10.0 6.7 132.0 200.0 134.0MediacostperTB 0 5.6 10.0 3.9 0.0 0.0 0.0Compatibility 15 7.6 6.4 10.0 114.0 96.0 150.0Transferrate 15 7.4 8.2 10.0 111.0 123.0 150.0Drivecost 0 1.3 10.0 1.1 0.0 0.0 0.0Vendoranalyses 15 4.5 7.1 10.0 67.5 106.5 150.0Scenariocost 35 2.2 10.0 1.8 77.0 350.0 63.0 Totalweightedscore 501.5 875.5 647.0

Conclusions and Recommendations for USGS Offline Archiving Requirements 15

6. Allarchivedfilesshouldbechecksummedandthechecksumstoredinthecorrespondinginven-toryrecord.WhenafileisretrievedfromeithertheSiloortheofflinemedia,integritycanbeverified.VerificationofeachretrievedfilemaynotbefeasiblebecauseofCPUimpacts.

7. Alldatashouldbemigratedtonewmediafrom3to5yearsafteritwaswritten.Althoughmosttapetechnologiescanreliablystoredataformuchlongerperiods,after5yearsthetransferratesanddensitiesthatoncewereleadingedgewillbecomeproblematic,anddriveswillbecomedifficulttomaintain.ThisisabestpracticesupportedbyNARA.

8. Whenwritingarchivetapes,thetapesshouldbeverifiedonaseconddrive.Thisverificationwillhelp

identifyanydriveincompatibility.Thispracticehasbeenimplementedandshouldcontinue.

9. Eachtimethisstudyisrevisited,thehighestscor-ingtechnologymaychange.ThischangedoesnotindicatethattheUSGSshouldchangeofflinetapetechnologiesfrequently.Stayingwithagiventech-nologyforseveralyearsisabenefit,evenifthetech-nologyisnottheleadingtechnologycontinuously.Thisstudyisasnapshotintime,andresultswoulddifferevenafewmonthsearlier/laterbecauseofnewhardwarereleases.TherecurrentlyisnocompellingreasontoabandonLTOtechnology.

10. TheUSGSshouldplantoupdatethistradestudyperiodically.Annuallymaybetoofrequenttoobservemarketchangesbecausedrivesaretypicallyupdatedona2-or3-yearcycle.

Appendix: Citations 17

Appendix: Citations

Vendor Sites

http://h18006.www1.hp.com/storage/tapestorage/tapedrives.html (Hewlett-Packard)

http://www.oracle.com/us/products/servers-storage/storage/tape-storage/index.htm (Oracle)

http://www-03.ibm.com/servers/storage/tape/index.html (International Business Machines)

http://www.quantum.com/Products/TapeDrives/Index.aspx (Quantum)

http://www.tandbergdata.com/us/en/products/drives/lto/ (Tandberg)

Consortium Sites

http://www.lto.org/newsite/index.html

Other

http://www.clipper.com/research/TCG2004040.pdf

http://www.redbooks.ibm.com/redbooks/pdfs/sg244632.pdf

http://www.computerworld.com/hardwaretopics/storage/story/0,10801,110667,00.html?source=NLT_SU&nid=110667

http://www.informationweek.com/news/showArticle.jhtml?articleID=187203674

http://www.techworld.com/storage/news/index.cfm?newsID=5888&pagtype=samechan

http://en.wikipedia.org/wiki/Holographic_Versatile_Disc

http://www.norsam.com/hdrosetta.htm

http://www.theregister.co.uk/2007/03/21/lto_beats_dlt/

https://www.bluestoragemedia.com/External/BigBlueBytes/Product%20Information/3592%20Gen%202/IBM%20System%20Storage%20TS1120%20Tape%20Drive%20Training%20Presentation.pdf

http://www.techworld.com/storage/features/index.cfm?featureid=3728

http://www.imation.com/euc/pdfs/EUC_07_Kenyon.pdf

http://www-03.ibm.com/systems/storage/tape/ts1130/index.html

http://h10010.www1.hp.com/wwpc/us/en/sm/WF25a/12169-304612-3446236-3446236-3446236-4150338.html

http://dlc.sun.com/pdf/316194802B/316194802B.pdf

http://www.computerworld.com/s/article/110667/Tape_drive_vendors_agree_on_encryption_technology?nid=110667&taxonomyId=019&source=NLT_SU

https://www.bluestoragemedia.com/External/BigBlueBytes/Product%20Information/3592%20Gen%202/IBM%20System%20Storage%20TS1120%20Tape%20Drive%20Training%20Presentation.pdf

http://www.tandbergdata.com/us/index.cfm/products/tape-drives/lto-drives/

http://www.norsam.com/pages/option_pages/index4.html


http://www.lto-technology.com/pdf/2006-7-25.pdf

http://itknowledgeexchange.techtarget.com/storage-soup/hp-quantum-in-cahoots-for-lto-5/

http://www.itjungle.com/fhs/fhs020508-story10.html

http://searchdatabackup.techtarget.com/news/article/0,289142,sid187_gci1355225,00.html

http://www-03.ibm.com/press/us/en/pressrelease/29245.wss

http://en.wikipedia.org/wiki/Linear_Tape-Open

http://www.infoworld.com/d/data-explosion/tape-dead-long-live-tape-090

http://www.engadget.com/2010/02/08/inphase-out-of-business-assets-seized-for-back-taxes/

http://h20195.www2.hp.com/v2/GetPDF.aspx/4AA0-7675ENW.pdf

ftp://service.boulder.ibm.com/storage/tape/clipper.pdf

http://searchdatabackup.techtarget.com/news/article/0,289142,sid187_gci1355225,00.html

Appendix: Citations 19

Publishing support provided by: Rolla and Lafayette Publishing Service Centers

For more information concerning this publication, contact: U.S. Geological Survey Earth Resources Observation and Science (EROS) Center 47914 252nd Street Sioux Falls, South Dakota 57198 (605) 594-6151

Or visit the EROS Center Web site at: World Wide Web: http://eros.usgs.gov/

Bodoh and others— A

rchive and Records Managem

ent—Fiscal Year 2010 O

ffline Archive M

edia Trade Study—Open-File Report 2010–1222

archive and records managementâ€”fiscal year 2010 offline archive

Documents