34
Riding Herd on Total Ownership Costs: Herding Sheep vs. Herding Cats Barry Boehm, USC STC 2015 Keynote Address October 13, 2015 2/21/2011 1

Riding Herd on Total Ownership Costs: Herding Sheep vs. Herding Cats Barry Boehm, USC STC 2015 Keynote Address October 13, 2015 2/21/20111

Embed Size (px)

Citation preview

Technical Tasks

Riding Herd on Total Ownership Costs: Herding Sheep vs. Herding Cats

Barry Boehm, USCSTC 2015 Keynote AddressOctober 13, 2015

2/21/201111Cowboys Herding Cats2/21/20112

OutlineSoftware O&M Often like Herding CatsSome Root Causes of Cat-Like BehaviorMany opportunities to reduce total ownership costa (TOC)By emphasizing software Changeability and DependabilityBoth rely on Maintainability via SERC System Qualities OntologyOpportunities organized via Maintainability Opportunity TreeAnticipate Modifiability NeedsDesign, Develop for Modifiability Anticipate Repairability NeedsDesign, Develop for RepairabilityExpedite Diagnosis Improve Modification and Repair Verifiability; SkillsConclusions

2/21/20113Problem and Opportunity (%O&M costs)

US Government IT: >75%; $62 Billion [GAO 2015]Hardware [Redman 2008]12% -- Missiles (average)60% -- Ships (average)78% -- Aircraft (F-16)84% -- Ground vehicles (Bradley)Software [Koskinen 2010]75-90% -- Business, Command-Control50-80% -- Complex platforms as above10-30% -- Simple embedded softwarePrimary current emphasis minimize acquisition costs2/21/201144Software O&M Often like Herding CatsSoftware and Users evolve in Incompatible directionsNon-Developmental Items (COTS, Clouds, Open Source)Independently evolving co-dependent external systemsMulti-mission sources of changeBreakage of brittle point-solution architecturesPriority changes: competition, technology, organizationsHerders are often ill-preparedMinimal voice in acquisitionMissing deliverables: diagnostics, test support, architecture documentation, CM supportDiversity of deliverables from multiple sourcesUnfamiliar domains, infrastructureMissing capabilities: Rainy-day use cases

2/21/20115Some Root Causes of Cat-Like BehaviorStovepipe acquisition of interoperating systemsIncompatible infrastructure, NDIs, user interfaces, Acquisitions based on lowest-cost, technically-acceptable implementation of fixed requirements, resulting inBrittle, point-solution architecturesCAIV-driven loss of information on post-delivery needsMinimal interpretation of technically acceptableJust implementing sunny-day requirementsMinimal maintainer participation, planning, preparationMissing maintainer deliverables: diagnostics, test support, architecture documentation, CM supportIncompatibilities among post-deployment evolution partiesInadequate SysE resources, leading to severe technical debtFirst to be impacted by optimistic budgets and schedules

2/21/20116The Conspiracy of Optimism Take the lower branch of the Cone of Uncertainty

USC-CSSE73/6/2011

706/25/07USC-CSSE8Added Cost of Minimal Software SysEBased on COCOMO II calibration data

06/25/07USC-CSSE9How Much Architecting is Enough?

Percent of Project Schedule Devoted to Initial Architecture and Risk Resolution

Added Schedule Devoted to Rework(COCOMO II RESL factor)

Total % Added Schedule10000KSLOC100 KSLOC10 KSLOCSweet SpotSweet Spot Drivers:Rapid Change: leftwardHigh Assurance: rightwardOutlineSoftware O&M Often like Herding CatsSome Root Causes of Cat-Like BehaviorMany opportunities to reduce total ownership costa (TOC)By emphasizing software Changeability and DependabilityBoth rely on Maintainability via SERC System Qualities OntologyOpportunities organized via Maintainability Opportunity TreeAnticipate Modifiability NeedsDesign, Develop for Modifiability Anticipate Repairability NeedsDesign, Develop for RepairabilityExpedite Diagnosis Improve Modification and Repair Verifiability; SkillsConclusions

2/21/201110Maintainability Challenges and ResponsesMaintainability supports both Dependability and ChangeabilityDistinguish between Repairability and ModifiabilityElaborate each via means-ends relationshipsMultiple definitions of ChangeabilityDistinguish between Product Quality and Quality in Use (ISO/IEC 25010)Provide mapping between Product Quality and Quality in Use viewpointsChangeability can be both external and internalDistinguish between Maintainability and AdaptabilityMany definitions of ResilienceDefine Resilience as a combination of Dependability and ChangeabilityVariability of Dependability and Changeability valuesOntology addresses sources of variationReferents (stakeholder values), States, Processes, Relations with other SQs Need to help stakeholders choose options, avoid pitfallsQualipedia, Synergies and Conflicts, Opportunity Trees

Product Quality View of ChangeabilityMIT Quality in Use View also valuableChangeability (PQ): Ability to become different productSwiss Army Knife, Brick Useful but not ChangeableChangeability (Q in Use): Ability to accommodate changes in useSwiss Army Knife does not change as a product but is ChangeableBrick is Changeable in Use (Can help build a wall or break a window?)

ChangeabilityAdaptabilityMaintainabilitySelf-DiagnosabilitySelf-ModifiabilitySelf-Testability

InternalExternalModifiabilityRepairabilityDefectsChangesTestabilityMeans to EndSubclass of

UnderstandabilityModularityScalabilityPortabilityDiagnosabilityAccessibilityRestorabilityABABC,DMeans to EndSubclass ofDependability, AvailabilityReliabilityMaintainabilityDefect FreedomSurvivabilityFault ToleranceRepairabilityTest Plans, CoverageTest Scenarios, DataTest Drivers, OraclesTest Software QualititiesTestabilityCompletePartialRobustnessSelf-RepairabilityGraceful DegradationChoices of Security,SafetyModifiabilityDependability, Changeability, and ResilienceTestability, Diagnosability, etc.06/01/2015USC-CSSEChangeabilityResilienceOutlineSoftware O&M Often like Herding CatsSome Root Causes of Cat-Like BehaviorMany opportunities to reduce total ownership costa (TOC)By emphasizing software Changeability and DependabilityBoth rely on Maintainability via SERC System Qualities OntologyOpportunities organized via Maintainability Opportunity TreeAnticipate Modifiability NeedsDesign, Develop for Modifiability Improve Modification V&VAnticipate Repairability NeedsDesign, Develop for RepairabilityImprove Repair DiagnosisImprove Repair, V&V ProcessesConclusions

2/21/20111408/04/201515Maintainability Opportunity Tree: ModifiabilityAnticipate Modifiability Needs Design/Develop for ModifiabilityImprove Modification V&VEvolution requirementsTrend analysisHotspot (change source) analysisModifier involvementSpare capacity; product line engineeringDomain-specific architecture in domain Regression test capabilities Address Potential ConflictsAddress Potential ConflictsAddress Potential Conflicts

Service-orientation; loose couplingModularize around hotspotsEnable user programmability

Modification compatibility analysis

Prioritize, Schedule Modifications, V&V

08/04/201516Maintainability Opportunity Tree: RepairabilityAnticipate Repairability Needs Design/Develop for RepairabilityImprove Repair DiagnosisRepair facility requirementsRepair-type trend analysisRepair skills analysisRepair personnel involvementReplacement parts logistics development Replacement accessibility deaignMultimedia diagnosis guidesFault localizationAddress Potential ConflictsRepair compatibility analysisRegression test capabilities Address Potential Conflicts Address Potential ConflictsImprove Repair, V&V ProcessesSmart system anomaly, trend analysisInformative error messagesReplacement parts trend analysisRepair facility design, developmentAddress Potential Conflicts

Switchable spare componentsPrioritize, Schedule Repairs, V&V Elaborating Modifiability Benefits - IEvolution RequirementsKeep, prioritize below-the-line IOC requirementsUse to determine modularization around sources of change, reduce ripple effects of changesTrend AnalysisIdentify, prioritize responses to sources of changeMarketplace, competition, usage trends, mobility trendsUse to refine, evolve architectureAgile Methods, User ProgrammabilityEnable rapid response to rapid changeHotspot AnalysisGather data on most common sources of changeUse to modularize architecture, reduce ripple effects of changes

2/21/201117183/27/2013 Use of Empirical Data in TOC Models:Pareto 80-20 Cost-to-fix Distribution Contracts: Fixed cost and nominal-case requirements; 90 days to PDR

18Rework Sources Analysis: Projects A and B- Change processing over 1 person-month = 152 person-hours3/27/2013 19CategoryProject AProject BExtra long messages3404+626+443+328+244= 5045Network failover2050+470+360+160= 3040Hardware-software interface620+200= 8201629+513+289+232+166= 2832Encryption algorithms1247+368= 1615Subcontractor interface1100+760+200= 2060GUI revision980+730+420+240+180 =2550Data compression algorithm910External applications interface770+330+200+160= 1460COTS upgrades540+380+190= 1110741+302+221+197= 1461Database restructure690+480+310+210+170= 1860Routing algorithms494+198= 692Diagnostic aids360477+318+184= 979TOTAL:136201353119C4ISR Project C: Architecting for ChangeUSAF/ESC-TRW CCPDS-R Project*3/27/2013 20When investments made in architecture, average time for change order becomes relatively stable over time* Walker Royce, Software Project Management: A Unified Framework. Addison-Wesley, 1998.

20Relative* Total Ownership Cost (TOC)For single system life cycle (TOC-SS)3/27/2013 21* Cumulative architecting and rework effort relative to initial development effort~5% architecture investment~5% architecture investment~25% architecture investment21Elaborating Modifiability Benefits IIand Repairability BenefitsService-Oriented Architecture improves InteroperabilityProduct-Line Engineering improves Total Ownership Cost (TOC)Identify, modularize around product line CommonalitiesDevelop domain architecture, interfaces to VariabilitiesFewer components to modify, repairImproved Repairability improves Availability, TOCAvailability = MTBF / (MTBF + MTTR)Stakeholder Value-Based V&V improves Cost, Mission EffectivenessPrioritizing inspection, test activitiesBalancing level of inspection, test activities vs. rapid fielding2/21/20112223Systems Product Line Flexibility Value ModelSystemsPL Flexibility Value Model

For Set of Products:Average Product CostAnnual Change CostOwnership TimePercent Mission-Unique, Adapted, ReusedRelative Cost of Developing for PL Flexibility via ReuseRelative Costs of ReuseAs Functions of # Products, # Years in Life Cycle:PL Total Ownership CostsPL Flexibility InvestmentPL Savings (ROI)2/21/201123Current Systems Product Line Model

2/21/2011242408/04/201525Reuse at HPs Queensferry Telecommunication Division

TimetoMarket

(months)Non-reuse ProjectReuse project2/21/201126Tradeoffs Among Cost, Schedule, and Reliability: COCOMO II

-- Cost/Schedule/RELY: pick any two points (RELY, MTBF (hours)) For 100-KSLOC set of featuresCan pick all three with 77-KSLOC set of features26Cost of Downtime SurveyIndustry SectorRevenue/HourEnergy$2.8 millionTelecommunications $2.0 millionManufacturing $1.6 millionFinancial Institutions $1.4 millionInformation Technology $1.3 millionInsurance $1.2 millionRetail $1.1 millionPharmaceuticals $1.0 millionBanking $996,000Source: IT Performance Engineering & Measurement Strategies: Quantifying Performance Loss, Meta Group, October 2000.

2/21/20112727Added TOC for Systems of Systems (SoSs)

Multi-mission sources of changeMany independently-evolving co-dependent systemsOver 155 on Future Combat SystemsMultiple-contract-change complexitiesAverage 141 vs. 48 workdays to process on 2 large SoSsLimited ability to apply interoperability improvements

These also affect systems that become parts of SoSs2/21/201128282/21/201129Examples of Utility Functions: Cost of DelayValueTimeReal-Time Control; Event SupportValueTimeMission Planning, Competitive Time-to-MarketValueTimeEvent Prediction- Weather; Software SizeValueTimeData Archiving

Critical Region292/21/201130Cost of Delay vs. Dependability Assurance- Early Startup: Risk due to low dependability - Commercial: Risk due to low dependability - High Finance: Risk due to low dependability - Risk due to market share erosion

COCOMO II: 012223454Added % test timeCOQUALMO: 1.0.475.24.1250.06P(L)Early Startup: .33.19.11.06.03S(L)Commercial: 1.0.56.32.18.10S(L)High Finance: 3.01.68.96.54.30S(L)Market Risk: .008.027.09.301.0REmSweet Spot

30Addressing Potential Conflicts With Performance: Loose vs. tight coupling (supercomputing) With Development Cost and Schedule: More to design, develop, V&V (rapid fielding)With Usability: Too many options (Office 2010)With Security: Too many entry points (Windows)With Scalability, Safety, Security: Agile methodsWith Dependability: User Programming, Self-AdaptivenessWith Interoperability: Multi-Domain ArchitecturesWith Cost, Resource Consumption: Spare Capacity

2/21/201131These are not always conflicts, but candidates to consider. Need to balance risk of too little Modifiability with risk of too much.

Example of Use: MS A Preparation GuidanceAlternatives analyzed shall include at least one architecture organized around:Common sources of life-cycle changeFor RPVs, these usually include user interface displays and controls, new payloads, self-defense and electronic warfare, data fusion, NCSoS protocols, and hardware maintainabilityRisk analysis and prototyping of critical off-nominal scenariosFor RPVs, these usually include communications outages, anti-RPV threats, noisy and intermittent data sources, redirected missions, and cross-RPV coordination of responsesAnalyses of alternatives shall include total ownership cost comparisonsBased on relevantly-calibrated life cycle cost models2/21/20113232OutlineSoftware O&M Often like Herding CatsSome Root Causes of Cat-Like BehaviorMany opportunities to reduce total ownership costa (TOC)By emphasizing software Changeability and DependabilityBoth rely on Maintainability via SERC System Qualities OntologyOpportunities organized via Maintainability Opportunity TreeAnticipate Modifiability NeedsDesign, Develop for Modifiability Improve Modification V&VAnticipate Repairability NeedsDesign, Develop for RepairabilityImprove Repair DiagnosisImprove Repair, V&V ProcessesConclusions

2/21/201133Conclusions Major opportunities to reduce Total Ownership CostsUS Government IT: $62 Billion O&M (over 75% of TOC)

Current investments overfocused on initial acquisition costsResulting in numerous herding-cats phenomena

Maintainabilty Opportunity Trees offer attractive means to reduce Total Ownership CostsBut need to balance risks of too little Maintainability with risks of too much 2/21/201134Chart818389159643231430681078402410214817653827714302555393237143347403600050505050

Percent of Time Added for Architecture and Risk ResolutionPercent of Time Added to Overall Schedule

Sheet1

Sheet118389159643231430681078402410214817653827714302555393237143347403600050505050

Percent of Time Added for Architecture and Risk ResolutionPercent of Time Added to Overall Schedule

Sheet2

Sweet Spot10000KSLOC100 KSLOC10 KSLOCPercent of Project Schedule Devoted to Initial Architecture and Risk Resolution

Added Schedule Devoted to Rework(COCOMO II RESL factor)

Total % Added Schedule

Sheet3

Chart15.466.136.667.338.394.354.895.315.856.693.824.294.665.135.873.824.294.665.135.873.824.294.665.135.87

(VL, 1)(L, 10)(N, 300)(H, 10K)(VH, 300K)Development Time (Months)Cost ($M)

Sheet1Development Time (Months)(VL, 1)18.25.4620.74.3524.33.8231.63.8238.93.82(L, 10)18.96.1321.44.8925.24.2932.84.2940.44.29(N, 300)19.46.66225.3125.94.6633.74.6641.44.66(H, 10K)207.3322.75.8526.75.1334.75.1342.75.13(VH, 300K)20.98.3923.76.6927.95.8736.35.8744.65.87

Sheet10000000000000000000000000

(VL, 1)(L, 10)(N, 300)(H, 10K)(VH, 300K)Development Time (Months)Cost ($M)

Sheet2

Sheet3

Chart10.0080.331.0083.0080.338130.0270.090250.2930.8250.117250.2660.7980.090.02640.16680.32040.11640.07680.23040.30.00750.32250.36750.30750.02250.067510.00181.0061.0181.00180.0060.018

Market Share ErosionEarly StartupCommercialHigh FinanceRELYRE = P(L) * S(L)Combined Risk Exposure

Chart20.0080.33130.3381.0083.0080.0270.090250.2660.7980.117250.2930.8250.090.02640.07680.23040.11640.16680.32040.30.00750.02250.06750.30750.32250.367510.00180.0060.0181.00181.0061.018

Market Share ErosionEarly StartupCommercialHigh FinanceRELYRE = P(L) * S(L)Combined Risk Exposure

Sheet1RELYMarket Share ErosionEarly StartupCommercialHigh FinanceVL0.0080.33130.3381.0083.008L0.0270.090250.2660.7980.117250.2930.825N0.090.02640.07680.23040.11640.16680.3204H0.30.00750.02250.06750.30750.32250.3675VH10.00180.0060.0181.00181.0061.018

Sheet2

Sheet3