AI m g p I ABSTRACT ll ee Background: The software market is becoming Results: The agile situation was investigated m S more dynamic which can be seen in frequent- through a series of case studies. The baseline oe ImplemenTIng leAn And AgIle n ly changing customer needs. Hence, software situation (plan-driven development) was eva- f TT SofTwARe developmenT In InduSTRy companies need to be able to quickly respond luated and the effect of the introduction of agile wI to these changes. For software companies this practices was captured, followed by an in-depth n Ag means that they have to become agile with the analysis of the new situation. Finally, a novel ap- R objective of developing features with very short proach, Software Process Improvement through l e e lead-time and of high quality. A consequence of the Lean Measurement (SPI-LEAM) method, was dA this challenge is the appearance of agile and lean introduced providing a comprehensive measu- en software development. Practices and principles rement approach supporting the company to v A of agile software development aim at increasing manage their work in process and capacity. SPI- e ln flexibility with regard to changing requirements. LEAM focuses on the overall process integrating od Lean software development aims at systemati- different dimensions (requirements, maintenance, p Kai Petersen cally identifying waste to focus all resources on testing, etc.). When undesired behavior is obser- m value adding activities. ved a drill-down analysis for the individual dimen- e n sions should be possible. Therefore, we provided T Objective: The objective of the thesis is to evalua- solutions for the main product development flow te the usefulness of agile practices in a large-scale and for software maintenance. The lean solutions I n industrial setting. In particular, with regard to agi- were evaluated through case studies. I le the goal is to understand the effect of migra- n ting from a plan-driven to an agile development Conclusion: With regard to agile we found that d approach. A positive effect would underline the the migration from plan-driven to agile develop- u S usefulness of agile practices. With regard to lean ment is beneficial. Due to the scaling of agile new T software development the goal is to propose no- challenges arise with the introduction. The lean R vel solutions inspired by lean manufacturing and practices introduced in this thesis were percei- y product development, and to evaluate their use- ved as useful by the practitioners. The practitio- fulness in further improving agile development. ners identified concrete decisions in which the lean solutions could support them. In addition, Method: The primary research method used the lean solutions allowed identifying concrete throughout the thesis is case study. As secon- improvement proposals to achieve a lean soft- dary methods for data collection a variety of ware process. approaches have been used, such as semi-struc- tured interviews, workshops, study of process documentation, and use of quantitative data. K a i P e t e r s e n Blekinge Institute of Technology Doctoral Dissertation Series No. 2010:04 2 0 School of Computing ISSN 1653-2090 1 0 2010:04 ISBN 978-91-7295-180-8 :04 Implementing Lean and Agile Software Development in Industry Kai Petersen Blekinge Institute of Technology Doctoral Dissertation Series No 2010:04 Implementing Lean and Agile Software Development in Industry Kai Petersen School of Computing Blekinge Institute of Technology SWEDEN © 2010 Kai Petersen School of Computing Publisher: Blekinge Institute of Technology Printed by Printfabriken, Karlskrona, Sweden 2010 ISBN 978-91-7295-180-8 Blekinge Institute of Technology Doctoral Dissertation Series ISSN 1653-2090 urn:nbn:se:bth-00465 Q: What are the most exciting, promising software engineering ideas or techniquesonthehorizon? A: I don’t think that the most promising ideas are on the horizon. They arealreadyhereandhavebeenforyears,butarenotbeingusedproperly. –DavidL.Parnas v vi Abstract Background: Thesoftwaremarketisbecomingmoredynamicwhichcanbeseenin frequently changing customer needs. Hence, software companies need to be able to quicklyrespondtothesechanges. Forsoftwarecompaniesthismeansthattheyhaveto become agile with the objective of developing features with very short lead-time and of high quality. A consequence of this challenge is the appearance of agile and lean softwaredevelopment. Practicesandprinciplesofagilesoftwaredevelopmentaimat increasing flexibility with regard to changing requirements. Lean software develop- ment aims at systematically identifying waste to focus all resources on value adding activities. Objective:Theobjectiveofthethesisistoevaluatetheusefulnessofagilepractices inalarge-scaleindustrialsetting. Inparticular, withregardtoagilethegoalistoun- derstandtheeffectofmigratingfromaplan-driventoanagiledevelopmentapproach. Apositiveeffectwouldunderlinetheusefulnessofagilepractices. Withregardtolean softwaredevelopmentthegoalistoproposenovelsolutionsinspiredbyleanmanufac- turingandproductdevelopment,andtoevaluatetheirusefulnessinfurtherimproving agiledevelopment. Method: The primary research method used throughout the thesis is case study. Assecondarymethodsfordatacollectionavarietyofapproacheshavebeenused,such assemi-structuredinterviews,workshops,studyofprocessdocumentation,anduseof quantitativedata. Results: Theagilesituationwasinvestigatedthroughaseriesofcasestudies. The baseline situation (plan-driven development) was evaluated and the effect of the in- troduction of agile practices was captured, followed by an in-depth analysis of the new situation. Finally, a novel approach, Software Process Improvement through the LeanMeasurement(SPI-LEAM)method, wasintroducedprovidingacomprehensive measurement approach supporting the company to manage their work in process and capacity. SPI-LEAM focuses on the overall process integrating different dimensions (requirements, maintenance, testing, etc.). When undesired behavior is observed a drill-down analysis for the individual dimensions should be possible. Therefore, we provided solutions for the main product development flow and for software mainte- nance. Theleansolutionswereevaluatedthroughcasestudies. Conclusion: With regard to agile we found that the migration from plan-driven to agile development is beneficial. Due to the scaling of agile new challenges arise with the introduction. The lean practices introduced in this thesis were perceived as usefulbythepractitioners. Thepractitionersidentifiedconcretedecisionsinwhichthe leansolutionscouldsupportthem. Inaddition,theleansolutionsallowedidentifying concreteimprovementproposalstoachievealeansoftwareprocess. vii viii Acknowledgements First and foremost, I would like to thank my supervisor and collaborator Professor Claes Wohlin for his support and feedback on my work, for the fruitful collaboration onpapers,andforalwaysrespondingtomyquestionsatanytimeofdayornight. I would also like to thank all colleagues at Ericsson AB for participating in my studies. Theseincludeparticipantsininterviews, workshops, andintheTiQanalysis team. Allofyouhaveprovidedvaluableinputandfeedbacktomyworkdespiteyour busy schedules, for which I am thankful. In particular, I also would like to express mygratitudetoEvaNilssonatEricssonforherstrongsupportandthusformakingthe implementations of the solutions presented in this thesis possible, and for providing importantfeedback. ThanksalsogotoPerOlofBengtssonforhiscontinuoussupport andcommitmentthroughoutmystudies. The support from my friends and colleagues from the SERL and DISL research groupsisalsohighlyappreciated. Inparticular,IwouldliketothankDejanBacaand ShahidMujtabafortheircollaborationonpapers,andforthegoodtimesafterwork. Lastbutnotleast,IwouldliketoexpressmysinceregratitudetomymotherFrauke foralwayssupportingmeinwhatIwantedtoachieveagainstallobstacles,andtomy brotherStephanforalwaysbeingthere. ThisworkwasfundedjointlybyEricssonABandtheKnowledgeFoundationinSwe- denunderaresearchgrantfortheproject“Blekinge-EngineeringSoftwareQualities (BESQ)”(http://www.bth.se/besq). ix