SOFTWA~RE COST NEGOTIA

Size: px
Start display at page:

Download "SOFTWA~RE COST NEGOTIA"

Transcription

1 1997 USC/CSE Annual Research Review SOFTWA~RE COST NEGOTA WARE COST OPTON ST March 10, 1997 Hoh n sunset. use-edu Center for Software Engineering Computer Science Department University of Southern California

2

3

4 - University of Southern California c s 1 E Center for Software Engineering The Problem Current Agreement: a $5M, 16-month upgrade to add Satellite Surveillance data services to a Mission Data ntegration Facility. New Rqmts: Add weather data services, to the MDlF upgrade. COCOMO Estimate: h t e Phaae~isribmion - m e WP Project File: <more>.../strkewfre.est Project Name: STRKEWRRE Modal File: <none> Deusloplnent bde: Embedded -6 New Requi $6.6M; '1 7.6 months Problem: How to resolve this cost issue in a collaboratively and effectively way?

5 - University of Southern California c Center for Software Engineering Solution Approaches A general capability to surface and negotiate conflicts and risks among requirements: WinWin* Capabilities to support the resolution of cost conflicts with functional and quality requirements based on early information - Aids for identifying cost conflicts with functional requirements: COCOMO* - Aids for identifying cost conflicts with quality requirements based on early information: QARCC* Capabilities to generate, visualize, and negotiate potential resolution options for cost conflicts: S-COST* *: info. is available at USC/CSE Homepage( ).

6 University of Southern California 1 c 1 E 1 Center for Software Engineering - How DOES S-COST WORK? ). t suggests options for resolving cost issues - Using Cost - Resolution Qption S- - Using Dption Strategy / Stakeholder ReWonships t provides visualization and negotiation aids for cost issueloption resolution

7 - a University of Southern California c E Center for Software Engineering Option Strategies Reduceldefer functionality Reduceldefer quality mprove tools, techniques. platform Relax schedule constraint 1 rnprove personnel capabilities Reuse software assets ncrease budget COCOMO KDS, DATA RELY, CPLX, TME TME, STOR, VRT, TURN, TOOL, MODP SCED ACAP, PCAP, AEXP, VEXP, LEXP, $WPM ADS, DM, CM, M Revised Win Condition Reduce OC cost, schedule Smaller produp to maintain Reduce cost, schedule, complexity Reduce slw cost, schedule mprove maintainability, other qualities Reduce cost if schedule was tight Reduce cost, schedule mprove quality Reduce cost, schedule May gain quality May enable product to reach competitive critical mass May increase RO Cons Capabilities unavailable to stakeholders Need to pay later if deferred Stakeholders lose quality capabilities (see QARCC table) ncrease tool, training, platform costs Reducing tool, platform experience would increase s/w cost Defer stakeholders' use of product capabilities Projects losing better people will suffer Potential stafing difficulties, delays ncreased cost/person-month unless low-cost outsourcing Stakeholders may lose quality capabilities Risk of overestimating reuse Added funds may not be available

8 A University of Southern California c E Center for Software Engineering L Reduceldefer functionality - COCOMO cost drivers: KDS, DATA - Pros: Reduce OC cost, schedule; Smaller product to maintain - Cons: Capabilities unavailable to stakeholders; Need to pay later if deferred mprove tools, techniques or platform - COCOMO cost drivers: TME, STOR, VRT, TURN, TOOL, MODP. - Pros: Reduce slw cost, schedule; lmprove maintainability, other qualities - Cons: ncrease tool, training, platform costs; Reducing tool, platform experience would increase s/w cost

9 University of Southern California Center for Software Engineering Stakeholder: ~ ~ ,. i General nteroperator User Developer Customer Maintainer i Public Cost- Reduction Option Strateqies: : Reduceldefer Relax mprove mprove Reuse increase i functionality, schedule personnel tools, software budget i 1 1 constraints capabilities platform assets a quality

10 University of Southern California Center for Software Engineering VSUALZATON FOR OPTON GENERATON Option Generation Aids D ROLE Name Beferring Functionality PRORTY [very High rtr 1. Target Cost (BK) ". BQTA NTEGRATON HHLL a 1 ZUY3 Estimated Cost (BK) Cost Difference (BK) Target Schedule (M) bs.00 Estimated Schedule (M) 1 Schedule Difference (M) i Total : 5379 l~egotiation ~ idl Operations: Split Resolution Strategies: Reduce/defer Functionality Pros & Cons Pros of this strategy are th - Reduce cost,0c,schedule - Smaller produce to maintai Cons of this strategy are th flpply!

11 -- University of Southern California Center for Software Engineering L VSUALZATON FOR OPTON GENERATON (CONT'D) Name Deferring Functionality 'get ost Total : 5379 Operat ions: / Resolution Strategies: Split,Merge,nsert,Delete,Accumulate,lndividual,Sort-by-Cost,Sort-by-Priority,mport,Export,SelectAll, Unsele :All

12

13 -& University of Southern California Center for Software Engineering S-COST Strengths: - Provides a more thorough set of candidate cost-resolution options and analysis of their pros and cons - Provides visualization and negotiation support for human users S-COST Limitations: - Generally cannot generate situation-specific options. Conclusion: semi-automated approach is stronger than a heavily manual approach or a heavily automated approach. - Exploits strengths, avoids limitations of automated aids