Seminar Paper by Lorenz Hartmann

Embed Size (px)

Citation preview

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    1/35

    submitted: April 2009

    by: Lorenz Hartmann20th of April, 1985Bad Schwalbach

    Student-ID: 1016673

    University MannheimLehrstuhl fr ABWL und Wirtschaftsinformatik

    D 68131 MannheimPhone: +49 621 1811691, Fax +49 621 1811692

    Internet:http://wifo1.bwl.uni-mannheim.de/

    I NVESTIGATING THE INDUSTRIALABILITY OF GLOBAL DISTRIBUTED SOFTWAREDEVELOPMENT PROJECTS

    Seminar paper

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    2/35

    Content

    1 Introduct ion and purpose ....................... ......... ......................... ....... ........................... . 1

    2 Bas ics of software eng ineer ing ................................ ................................ ..................... 2

    2.1 Key character i tics ................................ ................................ ................................ . 2

    2.2 Plan based approach ................................ ................................ ............................... 3

    2.3 Sof t are cr isis and paradig revisal................................ ................................ ....... 4

    3 Industr ial ized software development ........................... ..... ........................ ........ .......... 5

    3.1 Def inition of industr iali ed sof t are development................................ .................. 5

    3.2 A new sof tware engineer ing method: agile development ................................ ........ 8

    3.3 Implications to review SEthe global context ................................ ........................ 10

    4 Th e global perspect ive ................................ ................................ ............................... 11

    4.1 Global sof tware development ................................ ................................ ............... 11

    4.2 Problems of GSD ................................ ................................ ................................ . 12

    4.3 Approachesto tack le the problems of GSD ................................ .......................... 13

    4.3.1 Established approaches ................................ ................................ ............. 14

    4.3.2 Open source sof tware development (OSSD) ................................ ............. 17

    4.3.3 Tools for distr i bured collaborative development and howthey can

    change current development practice ................................ ................................ .... 19

    5 Discuss ion ................................ ................................ ................................ ................... 22

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    3/35

    F igures

    F igure 1: Waterfall modell (R oyce 1987) ................................ ................................ ....... 3

    F igure 2: Cost overunin projects (The Standish Group 1994) ................................ ......... 4 F igure 3: Impacts of distance {{28Carmel E. 2001}} ................................ .................. 12

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    4/35

    T ables

    Table 1: Spectrum of sof tware ar tifacts (Taubner 2005, K ilian-Kehr et al. 2007) ............. 6

    Table 2: Evaluation of collaboration tool categor ies (Hildenbrand 2008) ....................... 20

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    5/35

    Abbrev iat ions

    CSDP Collaborative sof tware development platforms

    GSD Global sof tware development IDE Integrated development environments

    OSSD Open source sof tware development

    SDP Sof tware development process

    SE Sof tware engineer ing

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    6/35

    Keywords

    Industr iali ation of sof tware development

    Global sof tware development Collaborative sof tware development platforms

    Agile development

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    7/35

    1 Introduct ion and purpose

    By investigating the industr ialability of global distr i buted sof tware development

    projects one is faced pr imar ily with the following problem set: To which amount is it

    possi ble to standardi e processes, automate tasks and employ reusable components insof tware development practice in the general and in the global case. To answer this

    question one hasto know what sof tware engineer ing1 (SE) exactly means, which k ind

    of sof tware development approaches are available and why/howthey are applicable to

    the var ious forms of sof tware products. F ur thermore it is impor tant to understand that

    matured knowledge about sof tware development in a collocated situation exists.

    However there is high uncer tainty how to adapt this knowledge to new trends and

    developments in a globali ed wor ld (Herbsleb, Moitra 2001). F or instance projects are

    increasingly character i ed by globally distr i buted projects teams. This leads to

    diff iculties in communication, control and coordination. (Carmel, Agarwal 2001)

    This paper will give a general introduction to sof tware engineer ing (cp. 2) and review

    the current state of industr iali ation in sof tware development practice (cp. 3). In the

    main par t of the paper the issue will be transferred to the global basis. Different dimensions which inf luence the current global sof tware development practice will get

    introduced (cp. 4.1). Especially the question how to deal with distance will be assessed

    by referr ing to case studies and new f indings of research (cp. 4.3.1). The reader will get to know why there is a strong emphasi e for standardi ation of processes in global

    distr i buted sof tware development. Additionally the reader will also learn that there are

    exper ts who try to change this tendency of standardi ation towards higher vitality in SE

    by using new communication methods (cp. 4.3.2, 4.3.3). Later it will be presented how

    global sof tware development might suppor t the tendency for fur ther industr iali ation in

    SE. There will be a f inal discussion about the current state, the limits and trends of

    industr iali ation in SE (cp. 5).

    .

    1 Sof tware engineer ing (SE) and sof tware development (SD) are used synonymously in the paper.

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    8/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 2

    2 Bas ics of software eng ineer ing

    2.1 Key c h aracter ist ics

    Sof tware engineer ing differs signif icantly from other engineer ing disci plines; this is not

    only becauseit is still a juvenescent science that star ted to develop from scratch just 50

    years ago. It also differentiates from others due to the extensive need for customer

    involvement and its attitude to require knowledge from var ious f ields, especially the

    computing- and application domain (Wal et al. 1993). Another difference is the lower

    degree of industr iali ation in contrast to other engineer ing disci plines. Some exper ts

    point out that SE still has indications of craf t manufactur ing (Janen 2005, p.284) and

    emphasi e the high knowledge intensity and the creative attitude of SE (Meyer 2006).In contrast to other engineer ing disci plines SDPs are character i ed by an empir ical and

    nonlinear proceeding (Williams, Cockburn 2003).But nonetheless it has to be regarded

    that for some sof tware products mass production could already be established (e.g.operating system, off ice applications, etc.). So obviously there are many aspects to take

    care of bylook ing at the current state of sof tware development. But before getting lost

    into detail, the basics about this topic should get introduced. Inthe following it will be

    descr i bed what sof tware engineer ing is and howit has developed.

    Sommerville def ines sof tware engineer ing as an engineer ing disci pline that is

    concerned with all aspects of sof tware production from the ear ly stage of system

    specif ication to maintaining the system af ter it has gone into use. (2007, p.7)

    According to his taxonomy it consists of four main activities. In the sof tware

    specif ication par t engineers and sof tware users def ine the requirements of the sof tware

    to be developed. The sof tware development phase includes the design of the

    sof tware/sof tware system and its implementation. This step is seamlessly linked with

    sof tware validation, which covers unit and system testing besides reinsur ing therequirements of the customer. In the sof tware evolution phase sof tware is adapted to

    customer and market requirement. Analysis as well as maintenance of the existing

    system is done. Obviously the phases are over lapping extensively and are not

    representing a ser ial process. They should be seen as an under lying gener ic framework

    of the main models of sof tware development which have evolved until today

    (Sommerville 2007)

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    9/35

    Error! Use the Home tab to apply berschrift 1 to the text that you want to appear here.Error! Use the Home tab to apply berschrift 1 to the text that you want to appear here. 3

    2.2 P lan based approach

    The first approaches which derived from software en ineerin research were plan based

    software processes. The most known model is the waterfall model. It is composed in

    several cascadin phases. In theory each phase is performed sequentially startin not before the previous phase has finished and documentation is delivered (Royce 1987).

    But practically the sta es show a hi h level of interdependency (Sommerville 2007). As

    an example desi n problems are found durin codin or requirements are identifieddurin the desi n phase. Therefore the SDP is done in a sequence of iterations rather

    than in theoretically developed linear flow (Sommerville 2007). This involves many

    drawbacks. To limit extendin costs and time consumption some phases in development

    are usually put on hold after some iterations (e. . requirement analysis, desi n phase).

    This mi ht lead to impractical systems since requirements could have chan ed in the

    meanwhile. The ma jor problem of the process is its inflexible portioninof the pro ject

    into distinct sta es. Commitments have to be made at an early sta e in process, which

    makes it difficult to respond to chan in requirements (Sommerville 2007, p.67) andlead to the problem that durin the maintenance phase previous processsta es have to

    be repeated. (Sommerville 2007) (Fae ri, Hanssen 2007).

    F igure 1: Waterfa ll mode ll (Royce 1987)

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    10/35

    Error! Use the Home tab to apply berschrift 1 to the text that you want to appear here.Error! Use the Home tab to apply berschrift 1 to the text that you want to appear here. 4

    2.3 Software crisis and paradigm reversa l

    Plan based software development was the state-of-the-art until the be innin of the

    1990s. But it could not overcome the dilemma between keepin cost and time exposure

    low and deliverin useful complete systems (Sommerville 2007). Therefore it led tovery poor success rates of software pro jects and the so called software crisis in the

    be innin of the 1990s. The Chaos Report shows that only 16.2% of the examined

    pro jects could be delivered on-time and in-bud et and especially in lar e software

    pro jects only 42% of the initially planned features and functions could be implemented

    (The Standish Group 1994).

    Parallel Gibbs refers to several studies and reports that the avera e softwaredevelopment overshoot is scheduled by half ; lar er pro jects enerally do worse. (1994, p.86). Evidently the plan based approaches with extensive documentation and a

    si nificant overhead in plannin , desi nin and coordination seemed not to be

    appropriate for most software development pro jectsany lon er. The needs had chan ed

    si nificantly. Software development had become an on oin process with shorter life

    cycles instead of a process which always starts from scratch (Sommerville 2007). Due

    to the fact that business needs were rapidly chan in there was a demand for fast

    product delivery and features which could be inte rated steadily into a runnin system(Janen 2005). The previous software development techniques were identified as

    inefficient processes and there was an increasin demand for lower software production

    and maintenance costs (Sommerville 2007). One proposition to address these new

    developments was to extent the de ree of industrialized software development which

    will be discussed into more detail in the followin para raph(cp. 3.1).

    15.50%

    31.50% 29.60%

    10.20% 8.80%4.40%

    0.00%5.00%

    10.00%15.00%20.00%25.00%30.00%35.00%

    Under20%

    21 -50%

    51 -100%

    101 -200%

    201 -400%

    Over400%

    F igure 2: Cost overun in projects ( T he Standish Group 1994)

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    11/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 5

    3 Industr ial ized software development

    3 .1 Def in it ion of industr ial ized software development

    Industr iali ation can be def ined as the spread of highly productive industr ial

    approaches in the production of goods and services. (Zwahr 2006, p.263) Greenf ield

    def ines the patterns of industr iali ation as the assembling of products out of

    components, the automation of rote or menial tasks, the creation of product lines and

    supply chains as well as the standardi ation of processes, architectures and packaging

    norms. (2004, p.17) Subsequently it will be presented which patterns of

    industr iali ation are transferable to SE. In doing so, the questions will be answered how

    and to which extend they have been adapted to SE. F ur thermore the limitations of industr iali ation in SE will be introduced.

    F ollowing the taxonomy of Greenf ield standardi ation might be seen as the key

    character istic of industr iali ation. In SE standardi ation was f irst applied with thedevelopment of high level programming languages like C , which have provided

    language level and machine-architecture independency (Stroustrup 1986). Another

    cr itical innovation to extent standardi ation was the introduction of the system-

    independent and future proof (Maidl 2005, p.283) TCP/IP protocol which has

    facilitated the communication between workstations and boosted the use of network

    services of companies and individuals (Taubner 2005, p.292). F ur thermore

    or gani ations and committees (ISO, IEEE, W3C) established industry standards for

    sof tware development (Maidl 2005), which have established best practices in many

    areas (Taubner 2005). These innovations favored that the dominating job-shop

    production, which provoked the sof tware cr isis with its ability of being slow and

    expensive (Greenf ield, Shor t 2004), got challenged. Industr ial production pr inci ples like

    speciali ation, division of labor and automation of process steps got into the focus of interest. Instead of creating a sof tware ar tifact from scratch, sof tware products were

    increasingly combined out of sof tware components of selected suppliers (Greenf ield,

    Shor t 2004). This approach, which is called component based sof tware engineer ing

    (CBSE), is descr i bed by Sommerville as a process of def ining, implementing and

    integrating or composing loosely coupled independent components into systems (

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    12/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 6

    2007, p.440). There are many advantages of developing sof tware in this industr iali ed

    manner :

    Allocation of extensive development costs of sof tware products on high volumes

    of customers (economy of scale) (Maidl 2005, Taubner 2005) Sof tware becomes a commodity which can be bought easily according to

    individual preferences (Sof tware as a service) (Carr 2005) Sof tware as a commodity leads to market concentration (e.g.: Microsof t, SAP)

    which has the positive effect of general accepted standards (e.g.: f ile formats)

    (Taubner 2005)and reduction of niche-suppliers (Maidl 2005) Higher dependability and less development costs because reused components are

    already specif ied, designed, implemented and validated in work ing systems

    (Sommerville 2007) R eduction of process r isk par ticular when relatively lar ge components such as

    subsystems can be reused (Sommerville 2007) The overall SDP is aimed to be accelerated (Sommerville 2007)

    However the pr inci ples of industr iali ation are not generally applicable to all k ind of

    sof tware products. Sof tware ar tifacts differ depending on the number of customers they

    are developed for, the product var iability and the related ability for

    standardi ation(Taubner 2005, K ilian-Kehr et al. 2007), which is indicated in table 1.Products Spe cific solutions

    Syst em a nd ma ss ma rk e t p roducts

    App lic a tion softw a r e

    Ent e r p ris esoftw a r e

    Int eg r a tion Indi vidu a l

    DMS, o pe r a tin gs yst em s ,app lic a tion s e r ve r or n e twork softw a r e, offic e

    and e-ma il

    port a ls , d a t aw a r eh ous e,CRM

    ERP , SCM,in ve ntor yma n ageme nt

    Introduction of ent e r p ris eapp lic a tions ,Int eg r a tion of co mp on ents and s pe cific softw a r e

    ex t ensions

    s a le s ord e r p roc e ss e s ,a irlin e b ookin gs yst em

    T able 1: Spectrum of software art ifacts ( T aubner 2005, K ilian-Ke h r et al. 2007)

    This table shows mass sof tware products like database management systems (DMS),

    operating systems, application server or network sof tware at the far lef t. They offer a

    high degree of common functionality (K ilian-Kehr et al. 2007), are widely used, have

    stable requirements and are not designed for a par ticular company. Thereforethey are

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    13/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 7

    well suited for being developed in components (K ilian-Kehr et al. 2007). The other

    categor ies to the r ight are character i ed by increasingly less oppor tunities for

    standardi ation. The extreme at the far r ight of the table is represented by individual

    customi ed sof tware solutions. F or these k inds of sof tware ar tifacts standardi ation is

    not meaningful as the requirements of the sof tware aretoo specif ic. Even if there would

    be components available the extra cost to f ind, understand and sometimes adapt the

    reusable component would be too high and the trade-off between ideal requirement and

    available component might not be found (Sommerville 2007).

    Obviously industr iali ation of SE is possi ble to a cer tain degree. Exper ts are even

    agreed that dur ing time many sof tware ar tifacts which belong to the categor ies on the

    r ight of the table will have the capability to be standardi ed and moveto the categor ies

    on the lef t of the table (Taubner 2005). However there are many implications of SEwhich will limit the level of industr iali ation and determine that a signif icant share of

    sof tware will not be a commodity within the foreseeable future (Taubner 2005, p.295).F ollowing ar guments suppor t this thesis:

    SE still has indications of craf t manufactur ing (Janen 2005, p.284)andit will

    take time to convince people to switch fromtheir established work ing procedures

    to more speciali ed and eff icient work ing methods (Janen 2005)

    According to McK insey only 7% of German IT employees work in the product business which has extensive potential for industr iali ation whereas 93% of the

    employees work in sof tware services (Hoch 2005) Component models are not as easy applicable as theory might illustrate. The

    components have to be independent and completely specif ied by their interfaces.

    There have to be component standards that facilitate the integration of components. F ur thermore middleware is needed that independent, distr i buted

    components are able to communicate with each other (Sommerville 2007).

    Besides these impor tant prerequisites the real burdens are even stronger. In SE thenot invented here syndrome (Janen 2005, p.284)is widely common and has

    lead to the failure many reuse-project. The reuse of components encounters high

    personal and emotional resistance (Janen 2005). This issue is even enforced

    when the source code of components is hidden dueto propr ietary r ights which is

    highly common in industry. Components of ten have to be seen as a black box,

    which decreasesthe trustwor thiness in these components because reliability and

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    14/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 8

    stability cannot be ensured (Sommerville 2007).Moreover the black box pr inci ple

    can lead to increased maintenance costs or even incompati bility when there is a

    system change (Sommerville 2007). Phases like R E and design are creative activities which are highly knowledge

    intensive (Meyer 2006). Par ts of SE offer less capability to be standardi ed and

    approaches haveto be foundto makethese processes more eff icient.

    Addressing the last topic of the previous enumeration agile development will be

    introducedin the next paragraph.

    3 .2 A new software eng ineer ing met h od: ag ile development

    The sof tware cr isis has not only led to a shif t towards a higher degree of industr iali ation of sof tware development; the static plan based work ing methods were

    also questioned. It was recogni ed that most SDPs differ in many cases fromthe pre-

    def ined process of other engineer ing disci plines that follow a cer tain assembling order

    and produce the same results every time (Schwaber, Beedle 2001). People got aware

    that most SDPs are character i ed by an empir ical and nonlinear proceeding (Williams,

    Cockburn 2003) and are marked by inevitable changes throughout its life cycle

    (Highsmith, Cockburn 2001, p. 120)like requirements changes andtechnology changes.

    In the mid 1990s practitioners f irst developed methodologies to face this issue andembrace rather than reject higher rates of change (Williams, Cockburn 2003, p.39)

    within the development process (see 2.3).F ollowing the tar get to facilitate the process

    of sof tware development and enhance quality, speed of delivery and customer

    satisfaction these approaches were called agile methods (Williams, Cockburn 2003,

    Sommerville 2007).

    The key character istics of these methods are the focus on incremental delivery: The

    system is developed from abstract specif ications and then iteratively ref ined by using customer input. Well known agile methods are Extreme Programming (XP) (Beck

    1999), Crystal (Highsmith, Cockburn 2001), Scrum (Schwaber,Beedle 2001) andF DD

    (De Luca 2009). In 2001the Agile Alliance, a group of the representatives of the

    leading agile methods formulated the four basis pr inci ples of agile methods:

    Individuals and interaction instead of processes andtools

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    15/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 9

    Executable sof tware instead of extensive documentation Customer collaboration over contract negotiation R esponding on change over following a plan

    According to the f irst pr inci ple programmer following the XP approach work in pairswhich leads to higher involvement, motivation (F owler 2006) and shows a similar

    productivity (due to less rework) astwo programmers work ing separately (Williams et

    al. 2000). The four th pr inci ple for instance implies that changes of system requirements

    have to be expected and that the system has to be designed to accommodate these

    changes (F owler, Highsmith 2001). The four generative rules of agile methods lead to

    higher customer collaboration/user engagement and the distr i bution of decision mak ing

    author ity over the whole development team (Highsmith, Cockburn 2001). Advantages

    of XP arethat the f inal product is more consistent with the business needsthan by using classic sof tware development methods (Sommerville 2007) andthe high applicability to

    turbulent, high change environments (Highsmith, Cockburn 2001). However in

    literature there dominates the opinion that agile methods are only suited to the

    development of small or medium-si ed business systems and cer tain personal computer

    products (e.g. computer games) (Williams, Cockburn 2003, Sommerville 2007).

    R easons for this limitation are as followed:

    Customer involvement, which can be a cr itical success factor for the SDP,depends on the willingness and availability of the r ight customers who can

    represent all system stakeholders (F aegr i, Hanssen 2007, Damian 2007). Through the attempt to deliver running sof tware quick ly and with a small amount

    of documentation there is strong emphasi e on extensive formal and informal

    communication and coordination dur ing the SDP. It incurs that programmersshould have suitable personalities for intense involvement and that the project

    should be character i ed by a low level of employee turnover (Sommerville 2007).

    Case studies have shown that projects can collapse when suppor ters of agilemethods in cr itical position of the project leave the project team (F aegr i, Hanssen

    2007). The missing documentation may make the source code diff icult to understand

    which might create problems in maintenance and validation (Sommerville 2007). There might occur contractual problems. Normal contracting methods are based

    on system specif ications which are def ined at the very beginning of the project.

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    16/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 10

    Using agile methods system requirements are specif ied only dur ing the

    development process and contracting according to system specif ications is not

    applicable (Sommerville 2007).

    Never theless agile methods have faced a high level of recognition because theyintroduced an alternative methodology to the static plan based development approaches

    (Williams, Cockburn 2003). Although they wereinitially developed to suppor t sof tware

    development in a collocated environment since 2003 researchers and practitioners try to

    f ind out possi bilities to blend selected agile practices with the regular practices.

    (Williams, Cockburn 2003, p.40,Boehm 2002) Some exper ts are even convinced that

    agile methods are an adequate approach for the sof tware development of lar ge scale

    system developments and projects character i ed by distr i buted development teams

    (Hildenbrand et al. 2008, Eckstein, Josuttis 2004). One ar gument is that agile methodswere usedin open source sof tware development (OSSD) projects successfully which are

    character i ed by distr i buted development teams (see 4.3.2).

    3 .3 Impl icat ions to rev iew SE t h e global context

    Summing up the implications of sof tware development, it should be mentioned that SE

    has changed signif icantly af ter the sof tware cr isis in the ear ly 1990s. Onthe one hand

    industr iali ation has comeinto the focus and the ideas of standardi ation, f lexi bility,and off-the-shelf products have become highly recogni ed (cp. 3.1). F or the f irst time

    sof tware products were split into modules which imply the oppor tunity to distr i bute

    SDPs not just according to tasks (R E, implementation, testing, etc.) like in plan based

    approaches. Using CBSE companies have received the chance to allocate the

    development of components of a sof tware system to different teams and let them work

    independently on a smaller piece of sof tware. Therefore sof tware can be developed in

    remote locations and simultaneously by different companies. This implicates higher

    eff iciency through division of labor and simultaneous work but it can lead to higher coordination and communication effor ts as well. These aspects get even more relevant

    in a global context (cp. 4.1, 4.2). But people recalled as well that the implementation

    par t of SE is in many cases an empir ical and nonlinear process (cp. 3.2). They

    introduced agile methods to make the implementation phaseitself more eff icient. In the

    main par t of the paper it will be investigated how these two trends in sof tware

    development are employed in the global context.

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    17/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 11

    4 Th e global perspect ive

    4 .1 Global software development

    In the last two decades sof tware development practice was forced to adapt to the

    ongoing globali ation of business. National markets turned into global markets and new

    forms of competition and cooperation across national borders hadto be developed. The

    result is global sof tware development (GSD) which evolved out of several reasons. The

    benef its can be summar i ed as followed (Herbsleb, Moitra 2001, Damian, Moitra 2006,

    Ye et al. 2007):

    Delegating development tasks to the most economically viable places to be more

    cost competitive which is seen asthe initial dr iving force of outsourcing Capitali e the global resource pool and recruit talented and suitable people for

    cer tain tasks regardless of location, nation andtime zone differences Tack le demand for faster product delivery by using time zone differences in

    round-the-clock development Exploit market oppor tunities of proximity to the market, including knowledge of

    customers andlocal conditions

    R egarding these goals it seems that the idea of the industr ialization of sof twaredevelopment which was introduced beforeis simply transferredto the global issue:

    Simultaneous engineer ing is extended by round-the-clock development. Instead of work ing with distr i buted teams within one country and between

    strongly related or ganizations, in GSD projects people are or ganizationally,

    spatially and time dispersed. F ollowing the division of labor pr inci ple employees work in vir tual teams for

    sometimes one single project by using information and communicationtechnologies.

    Never theless the practical exper iences have shownthat the established approaches and

    methods of SE are not fully transferable to GSD (cp. 4.2). In GSD many mechanism

    and rules are different. SE on a global basis is character ized by a higher distance in

    contrast to collocated sof tware development. This refers to its high distr i bution of

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    18/35

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    19/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 13

    tools to suppor t collaboration and the lack of informal communication (Herbsleb,

    Moitra 2001). Communication dimension: F ormal and informal communication is steadily used

    in sof tware development projects. F ormal communication is needed to apply

    crucial tasks like def ining interfaces, updating project status, determing

    responsi bilities. Informal communication is used because of the high vitality of a

    SDPs. It gets more relevant the more uncer tainty the project inhi bits (e.g.:

    requirements are not def ined at the beginning) (Damian 2007). Informal talks

    keep the people updated about what is happening aroundthem. Conveniently they

    get to know who is work ing at which topic, the status of sub-projects or which

    person has exper tise in which area. The absence of informal conversation, which

    is typically for projects at remote sites, can keep crucial problems unrecognizedand thereforelead to misalignment between concurrent work which causes rework

    (Damian, Zowghi 2003). Knowledge management dimension: Since there areless possi bilities for informal

    communication and moreover cultural problems (e.g.: fear to loss of intellectual

    proper ty) knowledge management in a GSD project has to not managed more

    intensively as in a collocated projects. F iltered information should be avoided and

    reuse oppor tunities should be given (Herbsleb, Moitra 2001).

    Technical dimension: In GSD it is needed to plan technical issues in detail at the beginning of the project. Otherwise crucial problems might ar ise dur ing the

    project through different versions of tools or incompati ble data formats. Moreover

    tasks like the transmission of cr itical data in conf iguration management have to be

    meticulously planned and executed in contrast to a non GSD (Herbsleb, Moitra2001). Eber t sees performance rapidly decreasing when multisite use is involved

    in cer tain tasks, dueto heterogeneous server and network infrastructure.(Eber t, De

    Neve 2001)

    F aced with these burdens different approaches for eff icient GSD were developed which

    will be discussedin detail in the next paragraph.

    4 .3 Approac h es to tackle t h e problems of GSD

    There are two main approaches of GSDto tack le the analysed problems. The f irst

    der ived out of a commercial back ground. In the mid 1990s some companies star ted as

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    20/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 14

    pioneers to develop sof tware globally. Using their exper ience they ref ined the way to

    deal with GSD. These established approaches of GSD will get introduced in the next

    paragraph by referr ing to exper iences from different cases related to GSD (cp. 4.3.1).

    The other approachis the application of non-commercial open source practices. These

    projects are also character ized by globally distr i buted developers but distinguish

    signif icantly from commercial approaches. Theyimply many other aspects and ideas

    and will be introduced subsequently (cp. 4.3.2). Later tools which suppor t distr i buted

    collaborative sof tware development will be presented. They havethe potential to reduce

    distance signif icantly and star t a new phase of GSD (cp. 4.3.3).

    4 .3 .1 Establ ish ed approac h es

    There aretwo main ideas howto deal with distance in GSD. Onthe one hand you cantry to avoid distance in the development process, on the other hand you cantry to

    handle distance more eff iciently that it becomesless relevant. Since the development of

    eff icient tools to suppor t collaboration was in its infancy in the beginning of GSD,

    pr imar ily tactics that went beyond communication technologies were applied in GSD

    aimed at reducing intensive collaboration, national and or ganizational cultural

    differences andtemporal differences (Carmel, Agarwal 2001, p.22).

    The most regarded issue to avoid distance is to reduceintensive collaboration, which istaken up in every case descr i ption regarded for this paper. Carmel states, that most

    companies try to engage the foreign entity only in tasks that are well def ined andstructured (Carmel, Agarwal 2001). The way how companies act is analogical to the

    predication of Carmel. Cusick for example suggests to have concept, analysis and

    design near ly 100% onshore andto hand structured tasks like construction and quality

    assurance over to the offshore par tner (Cusick, Al pana Prasad 2006).Cusick regards

    standardization of procedures as a key for a successful project. Having this clear

    separation of tasks these projects tend to follow a well structured waterfall approach.Agile, collaboration intensive methods are not or only rarely regarded (Eber t, De Neve

    2001, Cusick, Al pana Prasad 2006,Bur ger 2007, Kobitzsch et al. 2001, Battin et al.

    2001). Some exper ts ver ify this proceeding by stating that iterative and incremental

    development in distr i buted environments is diff icult (Paasivaara, Iassenius 2004).

    Another approachto limit the need for collaboration is to give the foreign entity full

    responsi bility for a system, system component, product or corporate process.F ollowing

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    21/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 15

    this CBSE based approachthe foreign entity is not using links with the center frequently

    and thus the ongoing collaboration is not as intense (Carmel, Agarwal 2001, Battin et al.

    2001). The under lying pr inci pal for this proceeding in GSD are the exper iences of the

    past. The company Alcatel studied projects for f ive years by distinguishing the degree

    of collocation. F or example they found out that collocated teams needed half of the time

    for defect detection in validation activities than a distr i buted team which works onthe

    same task (Eber t, De Neve 2001). This was ver if ied by Teasley, who repor ts that in

    collocated teams, productivity and job satisfaction are much higher (Teasley et al.

    2002). Therefore Eber t considers team-task collocation in a distr i buted project as a key

    objective. Teams that are assigned across several locations face many challenges that

    could impact their ability to work as a team (Eber t, De Neve 2001). These facts lead to

    the conclusion that the methodologies applied in GSD are in contrast to themethodologies applied in collocated sof tware engineer ing projects more focused on plan

    based sof tware engineer ing approaches. They favor clear ly separated tasks.

    Although this proceeding reduces distance, many of the pr ior named problems are not

    solved. To alleviate the remaining distance andto handle it more eff iciently several best

    practices have been developed since the beginning of GSD. A keylearning in most case

    studies is that there should be an established a relationshi p between team members of

    remote locations (Carmel, Agarwal 2001, Carr 2005). This way they can identify with

    the team, feel responsi bility for the team, are able to build up trust. As a result of that

    the electronic communication becomes effective. To addressthese f indings the concept

    of cultural liaison is widely used to br idge the cultural gap across sites and

    or ganizations (Carmel, Agarwal 2001, Carr 2005). According to this concept a key

    person of the project acts as ambassador andlinks the different sites of the project.Besides building up trust, the cultural liaison enables to spread domain exper tise

    because knowledge can be easily shared using this link (Carmel, Agarwal 2001, Carr

    2005). In a lar ge project of Motorola, key engineers of the remote locations moved to

    the main US based site of the company andtook par t in a three month workshop. They

    learned the system, hel ped completing R E and communicated this information back to

    their local teams. This measure was considered as a key success factor for the project by

    the company (Battin et al. 2001). Corresponding to this results Eber t claims that one

    way to improve is a heavy exchange of teams and management [] wh ich gradually

    build mutual understanding. (2001, p.68) Moreover he states that mixed teams should

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    22/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 16

    be set up to integrate diverse cultural and educational back grounds (Eber t, De Neve

    2001). Another central statement in many case studies is that the different development

    processes at the remote sites have to be managed eff iciently. Instead of installing

    common in-detail processes everywhereit is mostly advised to standardize on a higher

    level. Battin suggests that teams should not be forced into a common process because

    the learning curve of the team would be affected. Instead he proposesto handle the

    minor processes as a black box and concentrate on the interface def inition r ight in the

    initial architecture def inition phase. Pursuant to this the architecture of the project

    descr i bed byBattin followes three main pr inci ples: low coupling among network

    elements, well def ined interfaces and concise semantics for the network element

    behaviors (Battin et al. 2001). R egarding the problem of inconsistency in notations and

    terminologies between sites Battin suggests to agree on a set of common work products and common vocabulary at the beginning of the project. However regarding

    the implementation, the way how problems are solved differs in the examined case

    studies. Areas of dispar ity are for example the formation of teams, the allocation of

    work between teams, the way quality was ensured andthe decision in which phases

    iterative development should be applied. Never theless all papers point out the crucial

    need for communication tools. The commonly used tools are (Eber t, De Neve 2001,

    Cusick, Al pana Prasad 2006,Bur ger 2007, Kobitzsch et al. 2001, Battin et al. 2001,R ao

    et al. 2007):

    asynchronous messaging systems (e.g. e-mail, web forum) synchronous messaging systems (e.g. instant messaging, video-conferencing and

    telephone-conferencing, screen shar ing)

    and in most cases also:

    shared project repositor ies (for documents , source code etc)

    wik i webs (which substitute shared documents and glossar ies)

    Summar izing the observations of the different case studies the distance of GSDis still

    high and companies adapt their sof tware development practices to this situation.

    Companies switch fromtheir par tly agile development methods in collocated situations(Boehm 2002) to str ictly plan based approachesin global projects (Cusick, Al pana

    Prasad 2006). Most companies prefer to split work according to functional tasks and

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    23/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 17

    only some are following par tly a CBSE approach (Kobitzsch et al. 2001). In GSD

    coordination mechanisms like plans, schedules, system-level design and def ined

    processes are seen as a crucial success factor. They are considered to be moreimpor tant

    in geographically distr i buted sof tware development than in collocated sof tware

    development (Herbsleb, Gr inter 1999). But surpr isingly this f inding seems not to be a

    natural law. Open source sof tware development (OSSD), which is also character ized by

    geographically distr i buted development lacks many of the presented coordination

    mechanisms but also succeedsin developing sof tware of high quality and functionality

    (Mockus et al. 2002). Thereforeit will be presented in the following paragraph

    4 .3 .2 O pen source software development ( O SSD)

    OSSD is a way for building, deploying and sustaining lar ge sof tware systems on aglobal basis (Sommerville 2007). It is an alternative community-intensive socio-

    technical approach to develop sof tware systems, ar tifacts and social relationshi ps.

    (Scacchi 2007, p.464). One of the basic pr inci ples of this sof tware development style is

    that licenses and legal arrangements ensure that the source code of OSSDis generally

    available for remote access.Moreover OSSDtypically has a central body, consisting of

    some of the core developers, who are responsi ble for guiding the development of the

    project (Mockus et al. 2002). These basic arrangements ensure that the development

    process of OSSD projects is radically different to traditional SE projects:

    In many OSSD projects the systems are build by a lar ge number of volunteers(Mockus et al. 2002)

    The OSSD developers aretypically also end users (Mockus et al. 2002) Work is not assigned. The under lying belief in the freedom of choice ensuresthat

    developers just under take the tasks they areinterested in (Benk ler 2006) There is no explicit system-level design or even detailed design (Vixie 1999)

    There is no project plan, schedule or list of deliverables (Mockus et al. 2002)

    In OSSD projects sof tware informalisms (Scacchi 2002) take the place of the

    formalisms which is typical for traditional SE approaches. The most common types of

    informalisms used in OSSD include online discussion forums or threaded email

    messages. Some of the other waysto observe and par tici pate in project related topics are

    bulletin boards, group blogs as well as to do lists and project wik is (Scacchi 2007).

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    24/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 18

    These var ious types of sof tware informalism are accessi ble through project related

    websites and por tals. In the interplay with the informalism between different OSSD

    projects (e.g.: cross referencing) they form a web of informal, semi structured and

    processable information resources (Scacchi 2007, p.461). However OSSD do not have

    an anarchic attitude. Several mechanisms ensuregood sof tware development. Instead of

    explicit administrative control OSSD projects rely on gentle but suff icient social

    control (Scacchi 2007, p.462). OSSD projects are character ized by a sk ill-based

    mediocracy, in which project par tici pants self-or ganize around exper tise, reputation

    and accomplishments of core developers, secondary contr i butors and ter tiary reviewers

    as well as other per i pheral volunteers (Scacchi 2007, p.461). This includes voting on

    the approval of individual contr i bution to ongoing project sof tware (F ielding 1999),

    shared peer reviewing (Benk ler 2006, DiBona et al. 2005) or as well the projectsrecognition of a core developers status, reputation and geek fame (Scacchi 2007). This

    vir tual project management ensures to mobilize, coordinate, control, build and assure

    the quality of OSSD activities (Scacchi 2004). But the success of OSSD projects is also

    based onthe reduction of coordination effor ts. The vast major ity of source code (~80%)

    is created solely by core developers and most par tici pants typically contr i bute to just a

    single module (Scacchi 2007, p.460) of the system or tasks like defect repor ting

    (Mockus et al. 2002). The coordination concerns in the Apache OSSD project for

    instance are sharply limited by the stable asymmetr ically controlled interfaces. Anyfunctionality beyond [the basic development of the Apache server]is added by means of

    var ious ancillary projects that interact with Apache only through Apaches well def ined

    interfaces. (Mockus et al. 2002, p.342) This high interdependency of modules and

    components in a OSSD Project and the fact that some linchpin developers (Madey et al. 2005) work simultaneously in several OSSD projects implies the possi bility that

    OSSD projects are able to grow at a super-linear or exponential rate (Scacchi 2006,

    Schach et al. 2002) when modules or even sub systems are integrated into exciting

    OSSD projects (Schach et al. 2002).

    OSSD has manyimplications which might be useful for commercial GSD. One is that

    industr ialization in the sense of CBSE is applicable in global distr i buted projects. OSSD

    shows that coordination effor ts are limited by high modular ization and stable

    asymmetr ically controlled interfaces (Mockus et al. 2002). Moreover the evolutionary

    character istic of OSSD through the interrelations between projects might establish the

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    25/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 19

    view to see GSD more as an oppor tunity than as a burden. Although meanwhile shar ing

    and reuse became commonin traditional SE sof tware, the effor ts of traditional SE

    projects are still setup to produce systems whose growth and evolution is limited

    (Scacchi 2007). In contrast to that in OSSDthe focusis on sof tware evolution which is

    a process of co-evolution of interrelated and independent OSSD projects, people

    ar tifacts, tools code and project specif ic processes (Capiluppi, Michlmayr 2007, Weiss

    et al. 2006). The newtypes of socio-technical work practices, development processes

    and community network ing excel traditional SE (Scacchi 2007). F inally the extensive

    use of communication tools in OSSD already had and will have signif icant impact on

    SE to reduce distance fur thermore and facilitate GSD. Maurer is suppor ting this

    ar gument by stating that adequate and timely shar ing of information and knowledge in

    all directions, proactive change, management, and process monitor ing are some of theimpor tant factors required for successful project collaboration.(Maurer 1996) Therefore

    tools for distr i buted collaborative development will be discussedin the next paragraph.

    4 .3 .3 T ools for d istr ibuted collaborat ive development and h ow t h eycan c h ange current development pract ice

    The possi bility to handle distance more eff iciently by using powerful communication

    tools is used in commercial GSD but as it was observedin 4.3.2 even moreintensively

    in OSSD. The following paragraph investigates which features of communication toolsreduce distance signif icantly and enable the practicability of sof tware development

    similar to a collocated situation.

    In the regarded cases of commercial GSD developers of distr i buted projects employ two

    k inds of tools separately to conduct their work (cp. 4.3.1): On the one handthey use

    collaborative sof tware development tools (CSCW-tools) or also known as groupware

    and on the other hand they utilize integrated development environments (IDE).

    Groupwaretools hel p people to conduct a commontask and arethe basis for computer suppor ted cooperative work (Hildenbrand 2008). Some of the already presented tools

    like asynchronous messaging systems (e.g. e-mail, web forum) and synchronousmessaging systems (e.g. instant messaging, video-conferencing and telephone-

    conferencing, screen shar ing) are typical groupware applications (cp. 4.3.1). In contrast

    to that IDEs integrate individual sof tware development tools like auto complete

    functions, code generators and automatic code documentation (Sommerville 2007)

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    26/35

    Error! Use the Home tab to apply berschrift 1 to the text that you want to appear here.Error! Use the Home tab to apply berschrift 1 to the text that you want to appear here. 20

    which support the individual developer into a consistent interface and workin

    environment (Hildenbrand 2008). They are tailored to achieve better individual

    productivity but only rudimentarily support distributed collaboration within teams

    (Sommerville 2007). Examples for IDEs are open source solutions like NetBeans or

    Eclipse (Hildenbrand 2008). But obviously this separate use of roupware and IDE

    applications to conduct a certain task leads to inefficient development in GSD. In order

    to tackle this issue the open source community inte rated Groupware and IDE

    functionality into collaborative software development platforms (CSDP). Later these

    CSDP were successfully applied in several OSSD pro jects (Robbins 2005, Au ustin et

    al. 2002, Feller, Fitz erald 2002).

    T ab le 2: E va luation of co llaboration too l categories (Hildenbrand 2008)

    CSDP offer a central data repository (Robbins 2005), inte rated roupware

    functionality like IP telephony which is accessible either directly throu h the web

    browser or via IDE plu -ins and a fully internet based user interface (e. .sourcefor e.net). Hence CSDPs enable synchronous and asynchronous communication

    for a vertically continuous and collaboration intensive development process

    (Hildenbrand 2008, p.31) and the possibility for community buildin throu h the

    availability of various centralized information, artifacts sharin functions and

    stakeholder entities (Hildenbrand 2008). Moreover better traceability of individual

    process steps as well as relations between different software artifacts and contri butin

    stakeholders throu h hi h transparency can be achieved. In literature this is seen as a

    crucial success factor in GSD, because current and future PM [in GSD pro jects] will bemore concerned with trackin pro ject work processes and efficient and effective sharin

    of information and knowled e, amon pro ject contributors. (Chen et al. 2003, p.1)

    Due to the hi h performance of CSDP in OSSD pro jects the platforms were transferred

    to the commercial context. The most common CDSP are CodeBeamer, CollabNet

    Enterprise and Sour eFor e Enterprise. They offer a common set of CSDP features,

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    27/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 21

    have a considerable market share and [] have been used in multitude distr i buted

    projects. (Hildenbrand 2008, p.33) According to a study of R odr iguez as well as a

    rank ing of HildenbrandCodeBeamer was seen asthe best overall platform (Hildenbrand

    2008, R odr iguez et al. 2007). Although str ictly tool based SE approaches prevail

    signif icantly in terms of frequency and state of elaboration, in 2003 already 65% of

    sof tware development companies at least intended to use a CSDP for distr i buted

    projects by the year 2005 (Webster 2003).

    The different k ind of communication tools were presented here becausethey have the

    high potential to overcome the distance of GSD. Especially CSDP could make GSD

    similar to SE in a collocated situation. Due to the complete functionality they even

    could make agile sof tware development methods applicable in distr i buted collaborative

    sof tware development.

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    28/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 22

    5 Discuss ion

    The paper shows the most vital trends and developments in global sof tware

    development. It indicates that today the crucial question regarding SE is not about the possi bility of industr ialization in sof tware development because there is no doubt that SE has already reached a cer tain level of industr ialization. The question is what the

    (current) limits for industr ialization in SE are and howthey are inf luenced by GSD.

    As we have seen in paragraph 3 the level of industr ialization in SE has increased

    signif icantly until today. Numbers of indications of craf t manufactur ing in SE have

    been renewed by more professional approaches based on improved tools and

    automation, more appropr iate use of components and more eff icient project

    management. (Taubner 2005) Nonetheless industr ialization of SE solely cameinto the

    focus of interest due to the trend of outsourcing and offshor ing at the end of the 1990s

    (Taubner 2005). However it is crucial to understand that industr ialization was widely

    practiced already beforethis time. It was shown that already in the beginning of the

    1990s some sof tware ar tifacts were produced according to the pr inci ples of

    standardization, modular ization, economy of scale and a shif t towards commodities (cp.3.1). These pr inci ples were applied to increase the eff iciency by reducing cost/r isk and

    increase effectiveness by ensur ing sustainability (Maidl 2005). Although the ensuing trend to outsourcing/offshor ing at the end of the 1990s raised the attention of

    industr ialization of SE cur iously it countervailed the goal of increased eff iciency and

    effectiveness at f irst. As it was shown in the examined case studies companies could

    capitalize the global resource pool but were forced to reapply traditional plan based

    approaches of low eff iciency (cp. 4.3.1). This development signif ies the dilemma of

    industr ialization on the global basis. On the one hand pr inci ples of industr ialization like

    simultaneous engineer ing and division of labor can be put into a global context (cp. 4.1)

    but on the other hand oneis faced with additional distance. Or ganizational, spatial andtime dispersed vir tual teams lead to lower eff iciency of processes dueto diff iculties in

    communication, control and coordination (cp. 4.2). F or instance agile development

    methods which offer increased eff iciency in SE in collocated situations are mostly not

    applicable in a global context (cp. 4.3.1). However OSSD projects have shown that

    these burdens can be overcome especially by applying well-engineered communication

    platforms (cp. 4.3.2, 4.3.3). Hildebrand claims that only CSDP provide suff icient

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    29/35

    Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. Error! Usethe Hometab to apply berschr if t 1 to the text that you want to appear here. 23

    suppor t for spatial, temporal and or ganizational distr i bution [] and fea ture a superb

    integr ity and traceable information supply. (2008, p.36) Dueto these facts CSDP could

    decrease distance in GSD signif icantly and reducethe dilemma of industr ialization in a

    global context.

    To conclude shor tly: As already mentioned industr ialization has been established in SE

    to a signif icant degree. The increased amount of features of communicatin tools (cp.

    4.3.3) will be one reason whyindustr ialization of SE will even continue to expandin the

    future. Others are fur ther standandardzation, outsourcing and offshor ing which will

    encourage companies to strengthen their processes, be moreinnovative and manage

    tasks more eff iciently (Taubner 2005). The cr itical question for fur ther investigation is

    about upcoming limits for industr ialization of GSD specif ically and SEin general.

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    30/35

    R eferences

    Augustin, L., Bressler, D. & Smith, G. 2002, "Accelerating sof tware development through collaboration", Proceedings of the 24th International Conference onSoftware Engineering ACM New York, NY, USA, , pp. 559.

    Battin, R .D., Crocker,R ., Kreidler, J. & Subramanian, K. 2001, "Leveraging resourcesin global sof tware development", Software, IEEE, vol. 18, no. 2, pp. 70-77.

    Beck, K. 1999, "Embracing change with extreme programming", IEEE Computer, vol.32, no. 10, pp. 70-77.

    Benk ler, Y. 2006, The wealth of networks: How social production transforms marketsand freedom, Yale University Press, New Haven.

    Boehm, B. 2002, "Get ready for agile methods, with care", IEEE Computer, vol. 35, no.1; feasi ble and preferable in some circumstances, pp. 64-69.

    Bur ger, W. Offshoring and Outsourcing to INDIA 2007, .

    Capiluppi, A. &Michlmayr, M. 2007, "F romthe cathedral to the bazaar : An empir ical study of the lifecycle of volunteer community projects" in Open Source

    Development, Adoption and Innovation Spr inger, Boston, pp. 31-44.

    Carmel, E. & Agarwal, R . 2001, "Tactical approaches for alleviating distance in global sof tware development", Software, IEEE, vol. 18, no. 2, pp. 22-29.

    Carr, N.G. 2005, "Does sof tware matter - Sof tware-Industr ialisierung", Informatik-Spektrum, vol. 28, no. 4, pp. 271-273.

    Chen, F ., Nunamaker Jr., J.F ., R omano Jr., N.C. & Br iggs, R .O. 2003, "A collaborative project management architecture",System Sciences, 2003. Proceedings of the36th Annual Hawaii International Conference , pp. 12.

    Cusick, J. & Al pana Prasad 2006, "A Practical Management and Engineer ing Approachto OffshoreCollaboration", Software, IEEE, vol. 23, no. 5, pp. 20-29.

    Damian, D. & Zowghi, D. 2003, "R equirements engineer ing challenges in multi-sitesof tware development or ganizations", Requirements Engineering, vol. 8, no. 3,

    pp. 149-160.Damian, D. 2007, "Stakeholders in Global R equirements Engineer ing: Lessons Learned

    from Practice", Software, IEEE, vol. 24, no. 2, pp. 21-27.

    Damian, D. &Moitra, D. 2006, "Guest Editors' Introduction: Global Sof twareDevelopment: HowF ar Have WeCome?",Software, IEEE, vol. 23, no. 5, pp.17-19.

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    31/35

    De Luca, J. 2009, , Feature Driven Developemnt . Available: htt p://www.nebulon.com/fdd/index.html [2009, 21st of Apr il] .

    DiBona, C., Stone, M. & Cooper, D. 2005,Open sources 2.0, O'R eilly Media.

    Eber t, C. & De Neve, P. 2001, "Surviving global sof tware development", Software, IEEE, vol. 18, no. 2, pp. 62-69.

    Eckstein, J. & Josuttis, N. 2004, Agile Softwareentwicklung im Grossen: Ein Eintauchen in die Untiefen erfolgreicher Projekte, Dpunk t-Ver l., Heidel ber g,Germany.

    F aegr i, T.E. & Hanssen, G.K. 2007, "Collaboration, ProcessControl, and F ragility inEvolutionary Product Development", Software, IEEE, vol. 24, no. 3, pp. 96-104.

    F eller, J. & F itzgerald, B. 2002, Understanding open source software development,Addison-Wesley Longman Publishing Co., Inc.,Boston, MA, USA.

    F ielding, R .T. 1999, "Sharedleadershi p in the Apache project", Communications of the ACM, vol. 42, no. 4, pp. 42-43.

    F owler, M. 2006, 18th of July 2006-last update , Using an Agile Software Process withOffshore Development . Available: htt p://www.mar tinfowler.com/ar ticles/agileOffshore.html [2009, 21st of Apr il] .

    F owler, M. & Highsmith, J. 2001, "The Agile Manifesto", Software development, vol. 9,no. 8, pp. 28-35.

    Gi bbs, W.W. 1994, "Trendsin Computing: Sof tware's Chronic Cr isis",Scientific

    American, vol. 43, no. 9, pp. 86.

    Greenf ield, J. & Shor t, K. (eds) 2004,Software Factories: Assembling Applicationswith Patterns. Models, Frameworks and Tools. , Wiley Publishing, Inc.

    Herbsleb, J.D. & Gr inter, R .E. 1999, "Splitting the or ganization andintegrating thecode: Conway's law revisited", Software Engineering, 1999. Proceedings of the1999 International Conference on , pp. 85.

    Herbsleb, J.D. &Moitra, D. 2001, "Global sof tware development", Software, IEEE, vol.18, no. 2, pp. 16-20.

    Highsmith, J. &Cockburn, A. 2001, "Agile sof tware development: the business of innovation", IEEE Computer, vol. 34, no. 9, pp. 120-127.

    Hildenbrand, T. (ed) 2008, Improving traceability in distributed collaborative softwareengineering:a design science approach , Peter Lang, F rankfur t a.M.

    Hildenbrand, T., Geisser, M., Kude, T.,Bruch, D. & Acker, T. 2008, "AgileMethodologies for Distr i buted Collaborative Development of Enterpr ise

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    32/35

    Applications", International Conference on Complex, Intelligent and Software Intensive Systems, 2008 (CISIS 2008) , pp. 540.

    Hoch, D.J. 2005, "Gefahr Offshor ing - Sof tware-Industr ialisierung", Informatik-Spektrum, vol. 28, no. 4, pp. 287-291.

    Hofstede, G.J. 2005,Cultures and Organizations: Software for the Mind, McGraw-Hill.

    Janen, R . 2005, "Die Psychologie des Entwick lers - Sof tware Industr ialisierung", Informatik-Spektrum, vol. 28, no. 4, pp. 284-286.

    K ilian-Kehr, R ., Terzidis, O. & Voelz, D. 2007, "Industr ialisation of the sof twaresector", Wirtschaftsinformatik, vol. 49, no. 1, pp. 62-71.

    Kobitzsch, W., R ombach, D. &F eldmann,R .L. 2001, "Outsourcing in India [sof twaredevelopment]", Software, IEEE, vol. 18, no. 2, pp. 78-86.

    Madey, G., F reeh, V. & Tynan,R . 2005, "Modeling the F ree/Open Source sof twarecommunity: A quantitative investigation", Free/Open Source Software

    Development, , pp. 203-221.

    Maidl, J. 2005, "Spannungsfeld zwischen Standard und Prozessfhrerschaf t - Sof tware-Industr ialisierung", Informatik-Spektrum, vol. 28, no. 4, pp. 281-283.

    Maurer, F . 1996, "Work ing Group repor t on Computer Suppor t in Project Coordination", Project Coordination Workshop of the IEEE Fifth Workshops on

    Enabling Technologies: Infrastructure for Collaborative Enterprised (WET ICE) Stanford University, CA, USA, .

    Meyer, B. 2006, "The unspoken revolution in sof tware engineer ing", IEEE Computer,vol. 39, no. 1, pp. 121-124.

    Mockus, A.,F ielding, R . & Herbsleb, J.D. 2002, "Two case studies of open sourcesof tware development: Apache andMozilla", ACM Transactions on Software

    Engineering and Methodology, vol. 11, no. 3, pp. 309-346.

    Paasivaara,M. & Iassenius, C. 2004, "Using Interactive and Incremental ProcessesinGlobal Sof tware Development", Proc. ICSE 3rd Int'l Workshop Global Software

    Development IEEECS Press, , pp. 42.

    R ao, M.T., Ear ls, T.W. & Sanchez, G. 2007, "International Collaboration inTransor ganizational Systems Development: The Challenges of Global Insourcing", Journal of global information technology management, vol. 10, no.3, pp. 52.

    R obbins, J. 2005, "F ree/Open Source Processes and Tools, chapter Adopting OpenSource Sof tware Engineer ing (OSSE) Practices by Adopting OSSE Tools", , pp.245-264.

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    33/35

    R odr iguez, F ., Geisser, M., Berk ling, K. & Hildenbrand, T. 2007, "Evaluating Collaboration Platforms for Offshore Sof tware Development Scenar ios", L ecturenotes in computer science, vol. 4716, pp. 96.

    R oyce, W.W. Managing the development of large software systems: concepts and techniques , IEEEComputer Society Press,Monterey, California, United States1987, .

    Scacchi, W. 2007, "F ree/open source sof tware development: recent research results andmethods", Advances in Computers: Architectural Advances, vol. 69, pp. 243.

    Scacchi, W. 2006, "Understanding Open Source Sof tware Evolution", Software Evolution and Feedback, Theory and Practice, , pp. 181-206.

    Scacchi, W. 2004, "F ree/open source development practices in the computer gameindustry", Software, IEEE, vol. 20, pp. 59-67.

    Scacchi, W. 2002, "Understanding the requirements for developing open sourcesof twaresystems", IEE Proceedings-Software, vol. 149, no. 1, pp. 24-39.

    Schach, S.R ., Jin, B., Wr ight, D.R ., Heller, G.Z. & Offutt, A.J. 2002, "Maintainabilityof the Linux kernel", IEE Proceedings-Software, vol. 149, no. 1, pp. 18-23.

    Schwaber, K. &Beedle, M. 2001, Agile Software Development with Scrum, PrenticeHall PTR , Upper Saddle River, NJ, USA.

    Sommerville, I., 2007,Software engineering, Addison-Wesley, Har low, England; NewYork.

    Stroustrup, B. 1986, "TheC programming language", Addison-Wesley Series inComputer Science, .

    Taubner, D. 2005, "Sof tware Industr ialisierung", Informatik-Spektrum, vol. 28, no. 4, pp. 292-295.

    Teasley, S.D., Covi, L.A., Kr ishnan,M.S. & Olson, J.S. 2002, "R apid sof twaredevelopment through team collocation", Software Engineering, IEEE Transactions on, vol. 28, no. 7, pp. 671-683.

    The Standish Group 1994,The CHAOS Report , The Standish Group International Inc.,

    West Yarmouth, USA.Vixie, P. 1999, "Sof tware engineer ing" in Open sources: Voices from the open source

    revolution , eds. C. DiBona, S. Ockman &M. Stone, O'R eilly & Associates, Inc.Sebastopol, CA, USA, pp. 91-100.

    Walz, D.B., Elam, J.J. &Cur tis, B. 1993, "Inside a sof tware design team: knowledgeacquisition, shar ing, and integration", Commun.ACM, vol. 36, no. 10, pp. 63-77.

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    34/35

    Webster, M. 2003, "An end-user view of the collaborative sof tware development market", Market Research Report, IDC 30608, vol. 1.

    Weiss, M., Moroiu, G. & Zhao, P. 2006, "Evolution of open source communities" inOpen Source Systems, IFIP Spr inger, , pp. 21-32.

    Williams, L. &Cockburn, A. 2003, "Agile sof tware development: it's about feedback and change", IEEE Computer, vol. 36, no. 6, pp. 39-43.

    Williams, L., Kessler, R .R ., Cunningham, W. & Jeffr ies, R . 2000, "Strengthening theCase for Pair Programming", Software, IEEE, , pp. 19-25.

    Ye, Y., Nakakoji, K. & Yamamoto, Y. 2007, "R educing the Cost of Communicationand Coordination in Distr i buted Sof tware Development" in Software

    Engineering Approaches for Offshore and Outsources Development Spr inger, , pp. 152-169.

    Zwahr, A. 2006, "Brockhaus Enzyk lopdie in 30 Bnden", vol. 13.

  • 8/6/2019 Seminar Paper by Lorenz Hartmann

    35/35

    Eidesstattl ich e Erklrung

    Ich versichere, dass ich meine Di plomarbeit ohne Hilfe Dr itter und ohne Benutzung

    anderer als der angegebenen Quellen und Hilfsmittel angefer tigt und die den benutztenQuellen wr tlich oder inhaltlich entnommenen Stellen als solche kenntlich gemacht

    habe. Diese Arbeit hat in gleicher oder hnlicher F orm noch keiner Prfungsbehrde

    vor gelegen.

    Mannheim, den 12.May 2011

    Lorenz Har tmann