软体品质管理_第1页
软体品质管理_第2页
软体品质管理_第3页
软体品质管理_第4页
软体品质管理_第5页
已阅读5页,还剩52页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

2024/3/12软体品质管理IntroductionWhatissoftwarequality?Howcanitbemeasured?Howcanitbemeasuredbeforethesoftwareisdelivered?SomekeyqualityfactorsSomemeasurableindicatorsofsoftwarequalityIntroductionThinkofaneverydayobjecte.g.achairHowwouldyoumeasureit’s“quality”?constructionquality?(e.g.strengthofthejoints,…)aestheticvalue?(e.g.elegance,…)fitforpurpose?(e.g.comfortable,…)AllqualitymeasuresarerelativethereisnoabsolutescalewecansayAisbetterthanBbutitisusuallyhardtosayhowmuchbetterForsoftware:constructionquality(建造的品質)?softwareisnotmanufacturedaestheticvalue(美學上的價值)?butmostofthesoftwareisinvisibleaestheticvaluemattersfortheuserinterface,butisonlyamarginalconcernfitforpurpose?Needtounderstandthepurpose軟體品質因素MeasuringQualityTheQualityConcepts(abstractnotionsofqualityproperties)MeasurableQuantities(definesomemetrics)CountstakenfromDesignRepresentations(realizationofthemetrics)usabilityminutestakenforsomeusertask???timetakentolearnhowtouse?complexitycountprocedurecalls???informationflowbetweenmodules?reliabilityrunitandcountcrashesperhour???meantimetofailure?examples...FourKeyQualityConceptsReliabilitydesignermustbeabletopredicthowthesystemwillbehave:completeness-doesitdoeverythingitissupposedtodo?(e.g.handleallpossibleinputs)consistency-doesitalwaysbehaveasexpected?(e.g.repeatability)robustness-doesitbehavewellunderabnormalconditions?(e.g.resourcefailure)EfficiencyUseofresourcessuchasprocessortime,memory,networkbandwidthThisislessimportantthanreliabilityinmostcasesMaintainabilityHoweasywillitbetomodifyinthefuture?perfective,adaptive,correctiveUsabilityHoweasyisittouse?McCall’sQualityFactorsoperationrevisiontransitionCorrectnessreliabilityusabilityintegrityefficiencyMaintainabilityFlexibilityTestabilityPortabilityReusabilityInteroperabilityGeneral

utilityportabilityAs-isutilityMaintainabilityreliabilityefficiencyusabilitytestabilityunderstandabilitymodifiabilitydevice-independenceself-containednessaccuracycompletenessrobustness/integrityconsistencyaccountabilitydeviceefficiencyaccessibilitycommunicativenessself-descriptivenessstructurednessconcisenesslegibilityaugmentabilitySource:SeeBlum,1992,p176ProductoperationusabilityProductrevisionProducttransitionintegritymaintainabilitytestabilityreusabilityportabilityinteroperabilityoperabilitytrainingI/OvolumeAccesscontrolAccessauditStorageefficiencyconsistencyinstrumentationexpandabilitygeneralitySelf-descriptivenessmodularitymachineindependences/wsystemindependencecomms.commonalityefficiencycorrectnessreliabilityflexibilitycommunicatativenessI/OrateexecutionefficiencySource:SeevanVliet2000,pp111-3traceabilitycompletenessaccuracyerrortolerancesimplicityconcisenessdatacommonalityMeasurablePredictorsofQualitySimplicitythedesignmeetsitsobjectivesandhasnoextraembellishmentscanbemeasuredbylookingforitsconverse,complexity:controlflowcomplexity(numberofpathsthroughtheprogram)informationflowcomplexity(numberofdataitemsshared)namespacecomplexity(numberofdifferentidentifiersandoperators)Modularitydifferentconcernswithinthedesignhavebeenseparatedcanbemeasuredbylookingat:cohesion(howwellcomponentsofamodulegotogether)coupling(howmuchdifferentmoduleshavetocommunicate)QualityandhowtoachieveitProductquality–alwaysanissue

AfterWWII,industryintheUSandelsewherehassubstantiallyimprovedqualityviaextensivetestingandstatisticalqualitycontrolJapaneseindustryhasfollowedadifferentconcept,a.k.a.totalqualityinitiative(Deming,Juran),wherequalitycontrolisanintrinsicaspectoftheproductionprocess,notapost-productionactivityInotherwords,youshoulddesignthequalityproductandbuilditright,ratherthanbuilditandthenmakesureithasgoodqualitySEIandtheCapabilityMaturityModelSoftwarequalityimprovementhasledtotheestablishmentoftheSoftwareEngineeringInstituteatCarnegie-MellonUniversityCapabilityMaturityModel(CMM)aframeworktoassessthematuritylevelofanorganization’ssoftwaredevelopmentandmanagementprocessesCMMconsistsoffivelevelsofmaturityasmeasuredbyasetofguidelinescalledthekeyprocessareasHigherlevelsincreasecompetitiveness,reduceriskLevelsaremonotonic:levelnincludesallthecharacteristicsofallthelevelsbelow,n-1,n-2,etc.CapabilityMaturityModelLevel5OPTIMIZEDLevel4MANAGEDLevel3DEFINEDLevel2REPEATABLERISKCOMPETITIVENESSLevel1INITIALLevel1—Initialend…SoftwaredevelopmentfollowsnoprescribedprocessYoudon’tknowwhereyoustand,orwhenwillyoufinish,orwhatyouwillgetwhenyoufinishYouhavetobehappywithwhatyougetattheLevel2—RepeatableProjectmanagementprocessesandpracticesareestablished,inordertotrackprojectcosts,schedules,andfunctionalityStill,youjustpassivelytrackwhat’sgoingon…anddon’tknowexactlywhatyoushoulddoLevel3—DefinedAstandardsystemdevelopmentprocess(sometimescalleda“methodology”)purchasedordeveloped,andintegratedthroughouttheITunitoftheorganization…whichmeansthatthereisaprescribedsequenceofstepstofollow,plusyoumonitortheoutcomesLevel4—ManagedMeasurablegoalsforqualityandproductivityareestablishedConstantmonitoringgivesreal-timedataaboutthestatusoftheprocessandproductProcessparametersadjustedinordertoobtaindesiredoutcomes(thatismanagement,afterall)Level5—OptimizingThestandardizedsystemdevelopmentprocesscontinuouslymonitoredandimproved,basedonmeasuresanddataanalysisestablishedinLevel4Changescanbemadetoprocessparameters…Butwecanmakechangeintheprocessitself(e.g.,choosedifferentsequence,changepriorities,evenintroduceadditionalstepsordeletesomethatdonotcontributetotheprocess/productquality)CMMcertificationprocessCMMassessmentperformedbyqualifiedpersonnelfromSEI,throughinterviewsandanalysisofproceduresanddocumentsusedintheorganizationLevelscanbeawardedtoorganizationsorspecificprojects/projectgroupswithinanorganization.Eachlevelhasitsownquestionnaire,withanumberofmandatoryandoptionalquestionsInordertoqualifyforaspecificlevel,Yesanswersmustbepresentineachgroup(thoughminimumpercentagesaredifferent)CMMcertificationprocessEachquestionnairefocusesonanumberoftheso-calledKeyProcessAreas(KPAs)EachlevelofCMMscalehasanassociatedsetofKeyProcessAreas(KPAs)Toqualifyforaspecificlevel,issuesrelatedtoKPAsforthatlevelmustbeaddressedToprogressfromoneleveltothenextoneup,improvementsinappropriateKPAshavetobemadeLevelsaremonotonic–makesnosensetoconcentrateonKPAsfor,say,level4,ifthedevelopmentprocessisstillatlevel1or2關鍵流程領域(KeyProcessArea,KPA)階層特徵關鍵流程領域(KPA)品質方法5最佳層(Optimizing)持續流程改善最佳化的管理◎缺陷預防

◎技術變更管理

◎流程變更管理從流程變異最小化4管理層(Managed)產品與流程品質評量定量化的管理◎定量流程管理

◎軟體品質管理從預防問題發生3定義層(Defined)流程定義與制度化定性化的管理◎組織流程特徵

◎組織流程定義

◎同等審查

◎訓練課程

◎整合軟體管理

◎軟體產工程

◎群體間協調及早發現問題與矯正2重覆層(Repeatable)專案管理使用經驗的管理軟體專案追蹤◎軟體專案計畫

◎軟體副合約管理

◎軟體品質保證

◎軟體構型管理

◎需求管理專案最終的測試1初始層

(Initial)個人主義,毫無章法的管理SomeproblemswithCMM(atleastinitially)Focusesonproject/processmanagement,notproductdevelopmentIgnoresuseofcertainadvancedtechnologiesDidnotincorporateriskanalysisasaKPADidnotdefineitsdomainofapplicabilityButthingshaveevolved/diversified:ModelsthattheSEIiscurrentlyinvolvedindeveloping,expanding,ormaintaining:CapabilityMaturityModelIntegration(CMMISM)SW-CMMCapabilityMaturityModelforSoftwareP-CMMPeopleCapabilityMaturityModelSA-CMMSoftwareAcquisitionCapabilityMaturityModelSE-CMMSystemsEngineeringCapabilityMaturityModelIPD-CMMIntegratedProductDevelopmentCapabilityMaturityModelISO9000ISO9000seriesofqualitystandards,someofwhicharerelatedtosoftwareandISdevelopmentISOprinciples:SaywhatyouwilldoDoasyouhavesaidProvethatyoudidsoNOTE:thisisaverysimplifiedview–thecertificationprocessandthedocumentationrequiredtogothroughitarequiteextensiveanddetailedISOcertificationisperformedthroughnationalstandardsorganizationsCMMvs.ISOPartialoverlap,butdoubtsremainItseemspossiblethatanorganizationcanbecertifiedforISO9000andstillbeatlevel1oftheCMMscale,orbeatlevel3or4andstillunabletoobtainISO9000certification

Note:neitherISOnorCMMprescribeanyABSOLUTElevelofquality…buttheimprovementsinprocessmanagementarebelievedtohaveabeneficialeffectonqualitySQA:WhatisSQA?SQAistheprocessofassuringpeoplethateveryefforthasbeenmadetoensurethatsoftwareproductshavethedesiredlevelofreliability,maintainability,usability,andsalability.SQA是一種執行軟體評估與衡量的活動(BakerandFisher)Setofsystematicactivitiesprovidingevidenceoftheabilityofthesoftwareprocesstoproduceasoftwareproductthatisfittouse(G.SchulmeyerandJ.McManus,SoftwareQualityHandbook,PrenticeHall,1998.)SoftwareQualityAssuranceHowdoyouassurethequalityofyoursoftware?GoodprocessesGooddocumentationandartifactsAccountabilityLearnandimproveTheObjectivesofSQAMonitoringprocessesandproductsthroughoutthesoftwaredevelopmentlifecycletoensurethequalityofthedeliveredproduct(s)MonitoringtheprocessesProvidesmanagementwithobjective(客觀的)feedbackregardingprocesscompliance(承諾)toapprovedplans,procedures,standards,andanalysesMonitoringtheproductsFocusonthequalityofproductwithineachphaseoftheSDLCe.g.,requirements,testplan,architecture,etc.Objective:identifyandremovedefectsthroughoutthelifecycle,asearlyaspossibleSQA:AnSQAProgramMinimizingnumberofdefectsindelivereds/wCreatingmechanismsforcontrollingsoftwaredevelopmentandmaintenancesothatcostsandschedulescanbemetMakingcertainthatthedeliveredproductcanbeusedinitsintendedmarketplaceImprovingthequalityoffutureproductsSoftwaredefectsMistakesmadeatanypointinthesoftwareprocessRequirements,design,coding,maintenanceConsequencesInconvenience,lossofservice,financialloss,equipmentdamage,injury,deathThepercentageofdefectsfoundbyvariousmethodsPersonalcheckingofdesigndocumentsInformalgroupdesignreviewsFormaldesignspecificationsFormalcodeinspectionsModelingorprototypingPersonaldesk-checkingofcodeUnittesting(singleroutines)Functiontesting(relatedroutines)IntegrationtestingFieldtesting35%40%55%60%65%40%25%35%45%50%Thetruthofdefects1.Thelaterinthelifecyclethatanerrorisdetectedthemoreexpensiveitistorepair.2.Errorsremainlatentandarenotdetecteduntilwellafterthestageatwhichtheyaremade.54%oferrorsdetectedaftercodingandunittesting.45%oftheseerrorswererequirementsanddesignerrors.3.Therearenumerousrequirementserrors.Estimatesindicatethat56%ofallerrorsareerrorsduringtherequirementsstage. 4.Requirementserrorsaretypicallynon-clerical.RelativeCosttoRepairbasedonwhenitwasfoundRequirements-1timeDesign-3-6timesCode-10timesUnittest-15-40timesSystemtest-30-70timesFieldoperation-40-1000timesWhenshouldqualityassurancebedone?AteverystageinthesoftwareprocessTheProcessofSQA定義品質需求制定SQA計畫需求評估設計評估測試評估需求分析設計測試評估客戶滿意需求程度measurementfeedbackfeedback功能與完整性系統完整性與一致性執行效率與正確性SQAPlanningIEEEStd730-2002StandardforSoftwareQualityAssurancePlans12pagesIEEEGuideforSoftwareQualityAssurancePlanningdraftP730.287pagesSQAPlanningIEEESQAPPurpose(Section1oftheSQAP)Referencedocuments(Section2oftheSQAP)Management(Section3oftheSQAP)Documentation(Section4oftheSQAP)Standards,practices,conventions,andmetrics(Section5oftheSQAP)Reviewsandaudits(Section6oftheSQAP)Test(Section7oftheSQAP)Problemreportingandcorrectiveaction(Section8oftheSQAP)Tools,techniques,andmethodologies(Section9oftheSQAP)Codecontrol(Section10oftheSQAP)Mediacontrol(Section11oftheSQAP)Suppliercontrol(Section12oftheSQAP)Recordscollection,maintenance,andretention(Section13oftheSQAP)Training(Section14oftheSQAP)Riskmanagement(Section15oftheSQAP)ContentsofSQAPlan(sect1&2)PurposelistsoftwarecoveredstateportionofsoftwarelifecyclecoveredReferenceDocumentscompletelistofdocumentsreferencedelsewhereSect3-Managementorganization-depictstructureoforg.responsibilitiestaskstaskstobeperformedrelationshipbetweentasksandcheckpointssequenceoftasksresponsibilitiesofeachorganizationalunitSect4-Documentationidentify

requireddocumentsstatehowdocumentswillbeevaluatedminimumdocumentsSRS-SoftwareRequirementsSpecificationSDD-SoftwareDesignDescriptionSVVP–S.VerificationandValidationPlanSVVR-S.VerificationandValidationReportUserdocumentation-manual,guideSCMP–S.ConfigurationManagementPlanSect5-Standards,Practices,ConventionsandMetricsIdentifyS,P,C,andMtobeappliedHowcomplianceistobemonitoredandassuredMinimumdocumentationstandards,logicstructurestandards,codingstandards,testingstandardsselectedsqaproductandprocessmetricse,g,branch,decisionpointsSect6-ReviewsandAuditspurposedefinewhatreviews/auditswillbedonehowtheywillbeaccomplishedwhatfurtheractionsarerequiredMinimumSoftwareRequirementsReviewsPreliminaryDesignReviewevaluatetechnicaladequacyoftop-leveldesignMinSetofReviews/Audits(cont)CriticalDesignReviewacceptabilityofdetaileddesignsSoftwareVerificationandValidationPlanReviewadequacyofplannedverificationandvalidationFunctionalAuditallrequirementsinSRShavebeenmetPhysicalAuditsoftwareanddocumentsareconsistentandreadyIn-ProcessAuditManagerialReviewsSect7-TestAllteststhatarenotincludedinSVVPSect8-ProblemReportingPracticesandProceduresforreporting,tracking,andresolvingproblemsOrganizationalresponsibilitiesSect9-Tools,TechniquesandMethodologiesidentifythespecialsoftwaretools,techniquesandmethodologiespurposedescribeuseProcessAssessmentUseofstandardsandprocessmodelshasapositiveimpactonthequalityofthesoftwareproductDisciplined,controlleddevelopmentprocessExamplesinclude:ISO9001CMMCMUSEI,5levelsSPICE(ISO/IEC15504)SoftwareProcessImprovement&CapabilitydEterminationDevelopingastandardforsoftwareprocessassessmentISOjointcommittee,Europe,AustraliaIEEE1074,IEEE12207ProductAssessmentReviews,inspections,walkthroughsofPlans,reports,models,standardsProjectmanagement,qualityassurance,training,testplan(s)Requirements,analysis,architecture,detaileddesignmodel,testcasesIssueorproblemreportsMetricreportsTraceabilityreportsDocumentation,codingstandards…SoftwareReviewsTheymayincludemanagerialreviews,acquirer-supplierreviews,technicalreviews,inspections,walkthroughs,andaudits.Inspection:Aformalevaluationtechniqueinwhichanartifact(e.g.,softwarerequirements,design,orcode)isexaminedindetailbyapersonorgroupotherthantheoriginatordetectfaults,violationsofdevelopmentstandards,andotherproblems.reviewmembersarepeers(equals)ofthedesignerorprogrammer.dataiscollectedduringinspectionsforlateranalysisandtoassistinfutureinspections.ProblemReporting,Tracking,andResolvingDescribethepracticesandprocedurestobefollowedforreporting,tracking,andresolvingproblemsWhocanreportaproblem?Howisitreported?Howisittracked?Whodeterminesifitisaproblemthatgoingtoberesolved?Howisitassignedforresolution?Howdoesthepersonindicateithasbeencorrected?Whoreviewsittodetermineifitcanbeclosed

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论