Systematic review of success factors and barriers for software process improvement in global software development
2016; Institution of Engineering and Technology; Volume: 10; Issue: 5 Linguagem: Inglês
10.1049/iet-sen.2015.0038
ISSN1751-8814
Autores Tópico(s)Outsourcing and Supply Chain Management
ResumoIET SoftwareVolume 10, Issue 5 p. 125-135 Review ArticleOpen Access Systematic review of success factors and barriers for software process improvement in global software development Arif Ali Khan, Arif Ali Khan Department of Computer Science, City University of Hong Kong, Hong KongSearch for more papers by this authorJacky Keung, Corresponding Author Jacky Keung Jacky.Keung@cityu.edu.hk Department of Computer Science, City University of Hong Kong, Hong KongSearch for more papers by this author Arif Ali Khan, Arif Ali Khan Department of Computer Science, City University of Hong Kong, Hong KongSearch for more papers by this authorJacky Keung, Corresponding Author Jacky Keung Jacky.Keung@cityu.edu.hk Department of Computer Science, City University of Hong Kong, Hong KongSearch for more papers by this author First published: 01 October 2016 https://doi.org/10.1049/iet-sen.2015.0038Citations: 57AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract Nowadays, many software development organisations are globalising their work throughout the world. The motivation behind the software globalisation is competitive advantage and business profits that it yields to an organisation. Besides, these organisations are facing various challenges due to the distributed nature of the software development. However, software process improvement (SPI) is one of the biggest issues in the domain of global software development (GSD). The aim of this research study is to identify success factors and barriers to assist GSD organisations for successful implementation of SPI program. Accordingly, a systematic literature review approach was adopted to identify the success factors and barriers. A total of nine success factors and six barriers were identified that could impact SPI. Using the criteria of the factors having a frequency >50% as critical, total four success factors were ranked as critical success factors, i.e. management commitment, staff involvement, allocated resources and pilot projects. Moreover, total four barriers were classified as critical barriers, i.e. lack of resources, inexperienced staff, organisational politics and time pressure. 1 Introduction Software process improvement (SPI) is an approach used to comprehend the software process and to monitor the implementation of change in that process for the purpose of achieving the particular objectives such as to enhance software development period and budget along with the required efficiency [1]. Stelzer and Millis [2] described that SPI life cycle consists of four steps, i.e. PLAN (planning for process improvement), DO (operate the plan), CHECK (measuring the improvement) and ACT (implement the plan according to the given check). Different standards and models for SPI have been developed to move organisations towards optimising their software processes and also numerous instruments are developed to evaluate process proficiency. Capability maturity model integration (CMMI) [3] is among the SPI models that have been developed. It is a structured and systematic collection of best practices for process assessment and improvement. Moreover, the International Organisation for Standardisation (ISO) also embarked on the program to form a software process assessment standard. According to Viadiu et al. [4] ISO 9000 developed by ISO is a sequence of standards used to verify the quality of the system deployed by an organisation. Moreover, the ISO/IEC 15504 was developed by ISO for SPI under the software process improvement and capability determination (SPICE). SPICE was recognised as a project to come up with draft standards to test and create the awareness of the developing standards [4]. The research signifies that the effort devoted to these standards and models can help in developing a standard software system, decreasing time and cost, improving employee satisfaction and productivity [5-9]. SPI models like capability maturity model (CMM) [10], currently CMMI and standards like ISO's SPICE [4] emphasise on practices to accomplish quality software system. Emam and Koru [11] highlighted that slight attention has been paid to effectively implement these standards and models that have brought partial success to a number of SPI efforts. A study conducted by Niazi et al. [12] discussed that researchers show less intention towards an effective strategy to successfully implement the SPI standards. They were keen to have a proper guideline for 'how to implement' rather than 'what to implement'. There is a long tradition of SPI research in the fields of software engineering and information management systems [2, 5, 8, 13-17]. This research is motivated by an important gap found in SPI literature. Most of the previous research work discussed SPI in the domain of collocated software development [18], but recently most of the software development organisations have adopted global software development (GSD) in order to accomplish various benefits [19-21]. Babar and Niazi [13] shed light that term GSD is extending from emphasising only on cost reduction towards improvement of all the software development aspects. This change gives significant importance to process improvement program in distributed development. In general, the deployment of process improvement in GSD environment requires a long-term approach and need significant resources and time for complete implementation [13]. Ramasubbu [18] demonstrate that distributed team members operate beyond geographical boundaries where communication and coordination is more challenging and the need for implementation of SPI activities is much pronounced. Babar and Niazi [13] reported in their study that SPI team members in GSD face various problems such as developing SPI practices for distributed environments, building relationship between the distributed organisations, overcoming the temporal distance and addressing the cultural aspects. To effectively overcome these problems, the SPI team members need to have deep knowledge of designing and implementing different aspects of process improvement in the GSD domain [13]. Regardless of the significance of SPI implementation process, little investigation has been done in establishing models or techniques in order to efficiently deploy SPI activities in general and findings of success factors and barriers that can affect SPI implementation in a distributed environment in particular. Models and standards like CMMI, CMM, ISO 9000 and ISO/IEC 15504 provide adequate information for 'what activities to improve' but they did not provide a brief explanation 'how to improve' [14]. Organisations willing to deploy process improvement program need in-depth knowledge of success factors and barriers that can affect SPI implementation [13, 15, 18]. So far no study has been found that discussed success factors and barriers towards the effective execution of the process improvement program in the context of the distributed environment. In this study, we provide an in-depth overview of success factors and barriers that perform a significant role in the implementation of process improvement. Understanding of these factors and barriers can address the key area for SPI implementation. For this reason, the following research questions have been addressed. RQ1: What are the success factors and barriers that could affect SPI implementation in GSD? RQ2: Do the identified success factors and barriers differ from country to country? 2 Background study SPI offers organisations with a commanding means of evaluating their existing abilities for evolving software systems and accordingly they can identify their weaknesses and strengths. Zahran [22] defined SPI as 'the discipline of defining, characterising, improving and measuring software management, better product innovation, faster cycle time, greater product quality and reduced development costs simultaneously'. Several standards and models have been designed in the last several years, including, CMM [10], currently CMMI [3] and ISO's SPICE [4] for the purpose of enhancing software processes. These standards and models can lead the software organisations to develop quality products [15, 23]. Despite the availability of various process improvement standards, the success rate of SPI implementation is very low [13, 24]. Ngwenyama and Nielsen [25] reported that the failure rate of process improvement activities is 70% and one of the ground causes of this failure is the limited contribution paid towards the process improvement problems [13]. Niazi et al. [15] and Ramasubbu [18] highlighted that presently most of the software development organisations are globalising their work in order to compete globally and the practitioners need to know about the importance of SPI implementation in GSD domain. Nevertheless, the nature of process improvement problems is different in distributed environment [13]. Organisations should be attentive towards various challenges faced by SPI activities across geographical boundaries. According to Niazi et al. [15], the deployment of SPI in GSD context is more challenging as compared with collocated development, because of geographical, cultural and temporal distances among the development team members. Ramasubbu [18] discussed that most of the studies emphasised process improvement implementation in collocated software development. Prior research work on SPI has not examined the distributed nature of GSD organisations. GSD teams work in a distributed environment where the need for handling SPI implementation is much pronounced [18]. Overall, little empirical research has been conducted on efficient implementation of SPI programs in GSD and slight consideration has been given into success factors and barriers that are vital for successful execution of SPI activities [13, 15, 18]. The main aim of this work is to systematically extract all the available success factors and barriers that could impact the implementation of SPI in the domain of GSD. Various studies emphasised on the significance of key success factors and barriers. Those studies mentioned that the SPI staff should have a deep understanding of the factors that can bring a process improvement program towards successful deployment [13, 15, 18, 26]. It is also vital to evaluate the most critical factors which may have a strong influence on SPI activities in a GSD environment [15], as the literature shows that deployment of process improvement models like CMMI can be challenging because of limited resources and experience required for implementing CMMI practices [16]. In this study, we used systematic literature review (SLR) approach proposed by Kitchenham et al. [27] to extract success factors and barriers. The identification of success factors and barriers will decrease the failure rate as well as cost and time of process improvement implementation. 3 Research methodology The SLR approach has been adopted for classification of success factors and barriers of SPI implementation, whereas a similar method has been used by the authors in [24, 28, 29]. Khan [28] used SLR approach for the identification of critical success factors (CSFs) that can affect the contract management activities in distributed development. The same approach has been adopted by Khan and Azeem [29] for the classification of intercultural issues between the distributed teams. According to Kitchenham et al. [27], SLR is the systematic mode of reporting the outcomes extracted from the literature. SLR method offers a way of classifying, exploring and examining the present studies connected to any questions of interest and research areas [30]. Kitchenham [30, 31] classifies SLR into three main phases, i.e. planning the review, conducting the review and reporting the review. The SLR protocol is the outcome of the planning phase. Both authors of the article have participated in this SLR study. The first two phases (planning the review, conducting the review) of the SLR were performed by the first author. Second author has reported all the extracted data and properly document the results. The SLR protocol comprises various key phases as shown in Fig. 1. Fig. 1Open in figure viewerPowerPoint SLR protocol 3.1 Research problem Different studies discussed SPI in the context of collocated software development [18]. The distributed nature of SPI is ignored in previous research studies [13, 15, 18]. Generally, small-scale research has been done on the effective implementation of SPI in GSD and less attention has been given to the identification of success factors and barriers that could affect the SPI process [13, 15]. The main contribution of this work is to identify various success factors and barrier that could have a positive or negative impact on the implementation of the SPI program in the context of GSD. 3.2 Research questions The research questions for this study are discussed as follows: RQ1: What are the success factors and barriers that could affect SPI implementation in GSD? RQ2: Do the identified success factors and barriers differ from country to country? 3.3 Research process Digital databases were chosen based on prior research experience, references or suggestions provided by Chen et al. [32]. The databases chosen in this research study included publisher's site and index engines. Publisher's sites consist of published research study from their own database, while index engines contains a listing or index of data published at other publishing sites [32]. Table 1 discussed a list of selected publisher's and index sites. Table 1. Selected sources to be searched Publisher's sites Index engines Wiley Inter Science Inspec IEEE Xplore Cite Seer Science Direct Compendex ACM Digital Library Google Scholar Business Source Premier ISI Web of Science Springer Link The search strings used in these databases are based on the major keywords and alternative words of research questions. The keywords and their alternatives are chosen on the basis of available literature in the context of SPI and GSD [24, 28, 33, 34]. List of major keywords and their alternatives are discussed in Table 2. Table 2. Major keywords and alternative words Major keywords Alternative words factors factors, aspects, items, elements, variables barriers barriers, obstacles, hurdles, difficulties, impediments, hindrance SPI SPI, software process improvement, software process enhancement, CMM, CMMI, SPICE, software process enrichment, software process evaluation, software process assessment, software process appraisal GSD GSD, global software development, global software engineering, distributed software development, software outsourcing, offshore software development These keywords and their alternative words are concatenated using Boolean 'OR' and 'AND' operators in order to develop the search strings. All the selected databases were searched using the following search strings: ('factors' OR 'aspects' OR 'items' OR 'elements' OR 'variables') AND ('barriers' OR 'obstacles' OR 'hurdles' OR 'difficulties' OR 'impediments' OR 'hindrance') AND ('SPI' OR 'software process improvement' OR 'software process enhancement' OR 'CMM' OR 'CMMI' OR 'SPICE' OR 'software process enrichment' OR' software process evaluation' OR 'software process assessment' OR 'software process appraisal') AND ('GSD' OR 'global software development' OR 'global software engineering' OR 'distributed software development' OR 'software outsourcing' OR 'Offshore software development'). 3.4 Publications selection The exclusion and inclusion criteria are defined as below. 3.4.1 Inclusion criteria Those studies were considered that focus on SPI activities in GSD. The selected articles must be available in English and full text articles. These articles must be conferences, journals, workshops and books articles. Those studies definitely have more weight that provided empirical evaluations supported with case studies. It was not about rating any work, in fact it has been attempted to find the relevance of the work according to the proposed domain. 3.4.2 Exclusion criteria The studies that did not explicitly discussed the SPI success factors, barriers and models or framework have been excluded. Further exclusions of those studies have been made that did not provide significant details of SPI standards and models. The exclusion criteria also comprised of eliminating the duplicate results of articles. 3.4.3 Primary studies selection During primary studies selection process, various research articles were found and tollgate approach proposed by Afzal et al. [35] was used to refine the selection process. Tollgate approach comprised of five phases and is shown in Fig. 2 and described in Table 3. Fig. 2Open in figure viewerPowerPoint Tollgate approach for articles selection Table 3. Total selected articles using tollgate approach E-databases Search through search terms Inclusion/exclusion on the basis of title and abstract Inclusion/exclusion on the basis of introduction and conclusions Inclusion/exclusion on the basis of full text Total selected articles for primary study Percentage of final selcetd articles (n = 33) Wiley Inter Science 16 13 8 6 3 9.09 IEEE Xplore 20 15 13 11 8 24.24 Inspec and Compendex 31 25 16 11 0 0 Science Direct 25 18 16 13 10 30.30 Cite Seer 14 12 9 7 3 9.09 ACM Digital Library 22 18 11 7 4 12.12 Google Scholar 29 23 13 6 0 0 Springer Link 12 10 8 6 5 15.15 ISI Web of Science 17 14 10 3 0 0 Business Source Premier 35 22 14 8 0 0 total 221 173 113 71 33 100 In the first phase of tollgate approach, a total of 221 articles were selected from all the databases. Articles selection was based on given inclusion criteria. In the second phase, a total of 173 papers were selected after eliminating the duplicated articles. In third phase, only those articles were selected whose title and abstract involved keywords related to the proposed research questions. After applying the third phase, total 113 articles were sorted out. In fourth phase, the inclusion or exclusion was based on the 'introduction and conclusion'. Only those articles were selected which had an empirical background related to SPI and generally related to GSD. After applying this phase, total 71 articles were filtered and preceded to the final phase. During the final phase, the full text articles selection was done. In this phase, only those articles were selected, which were mainly related to the success factors and barriers of process improvement. A total of 33 articles were extracted as a primary study for this research. The results demonstrate that most of the articles related to SPI success factors and barriers are extracted from science direct (30.30%). Science direct is the most common and active digital library to publish articles associated to success factors and barriers. The list of the selected articles is mentioned in Appendix. 3.5 Study quality assessment The study quality assessment criteria were designed to assess the quality of the selected studies. The quality assessment was performed on the basis of the checklist provided in Table 4. The checklist was developed by following the given procedure in [24, 36, 37]. Table 4. Quality assessment checklist QA questions Score Description QA1: Is the scope of the study clearly discussed? QA2: Is the SPI technology used distinctly defined? QA3: Are the extracted results can describe the research questions? QA4: Does the study explicitly focused on GSD? QA5: Does the study discussed SPI standards and models? QA6: Does the adopted research method appropriate to answer the research questions? 1 studies that have explicitly fulfill the quality assessment research questions 0.5 studies that provides partial information concerning the quality assessment research questions 0 no information Those studies which contained answers of the research questions mentioned in the quality checklist are considered as ' yes' and one points were given. Similarly, if a study contained partial information regarding the research questions, then 0.5 points were awarded for that study. The studies that have not answered the questions were awarded with 0 points. After applying the above quality assessment criteria, total 33 articles were included in the final list of the selected primary studies. We also assess the selected articles through external review, which illustrate that these articles have sufficient quality to consider in this systematic study. 3.6 Data extraction The first author has listed all the identified success factors and barriers that have a positive or negative effect on SPI implementation. Second author has considered the frequency and percentage of each success factor and barrier in the selected articles. Second author extracted title of each primary study and publisher of each study. Complete list of all the selected 33 primary studies is discussed in Appendix. 3.7 Data synthesis Second author performed the data synthesis process. In this phase, all the extracted results were brought together and each question was exclusively evaluated against the results. From selected 33 articles, total of nine success factors and six barriers were identified as shown in Tables 5 and 6, respectively. Table 5. Identified success factors Sl. No Success factors Frequency (n = 33) Percentage References 1 management commitment 27 81 [LT1, LT2, LT3, LT4, LT5, LT6, LT7, LT8, LT9, LT10, LT11, LT12, LT13, LT14, LT15, LT17, LT18, LT19, LT21, LT23, LT24, LT25, LT26, LT28, LT29, LT31, LT32] 2 staff involvement 21 63 [LT1, LT2, LT3, LT4, LT5, LT8, LT9, LT11, LT12, LT14, LT15, LT17, LT1, LT19, LT21, LT23, LT24, LT26, LT28, LT29, LT31] 3 allocated resources 21 63 [LT1, LT2, LT3, LT4, LT5, LT7, LT9, LT10, LT11, LT12, LT13, LT14, LT15, LT18, LT19, LT21, LT23, LT24, LT26, LT28, LT31] 4 pilot projects 19 57 [LT1, LT2, LT3, LT4, LT5, LT6, LT8, LT10, LT12, LT14, LT15, LT17, LT18, LT21, LT24, LT26, LT28, LT29, LT31] 5 setting SPI goals 16 48 [LT2, LT3, LT4, LT5, LT9, LT11, LT12, LT14, LT15, LT17, LT18, LT19, LT21, LT23, LT29, LT31] 6 team training 13 39 [LT2, LT3, LT4, LT5, LT9, LT10, LT11, LT14, LT21, LT23, LT26, LT29, LT31] 7 information sharing 12 36 [LT1, LT2, LT3, LT4, LT5, LT7, LT9, LT10, LT15, LT17, LT19, LT28] 8 strong relationship 9 27 [LT1, LT2, LT7, LT10, LT12, LT14, LT26, LT28, LT31] 9 SPI awareness 6 18 [LT2, LT3, LT9, LT17, LT21, LT31] Table 6. Identified barriers Sl. No Barriers Frequency (n = 33) Percentage References 1 lack of resources 31 93 [LT2, LT3, LT4, LT5, LT6, LT7, LT8, LT9, LT10, LT11, LT12, LT13, LT14, LT15, LT16, LT17, LT18, LT20, LT21, LT22, LT23, LT24, LT25, LT26, LT27, LT28, LT29, LT30, LT31, LT32, LT33] 2 inexperienced staff 29 87 [LT2, LT3, LT4, LT5, LT6, LT8, LT9, LT10, LT11, LT12, LT13, LT14, LT15, LT16, LT17, LT18, LT19, LT20, LT22, LT24, LT25, LT26, LT27, LT28, LT29, LT30, LT31, LT32, LT33] 3 organisational politics 23 69 [LT5, LT6, LT8, LT9, LT10, LT11, LT12, LT13, LT14, LT16, LT18, LT19, LT20, LT22, LT23, LT25, LT26, LT27, LT28, LT29, LT30, LT31, LT32] 4 time pressure 17 51 [LT5, LT6, LT9, LT11, LT13, LT14, LT16, LT20, LT22, LT25, LT26, LT27, LT28, LT29, LT31, LT32, LT33] 5 staff turnover 14 42 [LT6, LT9, LT11, LT13, LT16, LT18, LT20, LT25, LT26, LT27, LT28, LT30, LT31, LT32] 6 lack of formal SPI implementation methodology 13 39 [LT9, LT11, LT13, LT16, LT20, LT25, LT26, LT27, LT28, LT29, LT30, LT31, LT32] 4 Results This section discussed results and analysis for each research question as mentioned in Section 3.2. 4.1 Identified success factors using SLR To answer RQ1, Tables 5 and 6 presents a list of identified success factors and barriers, respectively. The core objective for RQ1 is to extract all those success factors and barriers that can affect the SPI implementation. Table 5 and Fig. 3 illustrate the identified success factors along with their frequencies in the selected articles for primary study. Fig. 3Open in figure viewerPowerPoint Frequency distribution of identified success factors In Table 5, (n = 33) is the total number of selected articles for primary study and these articles are cited in Appendix. In this section, also the association is discussed between independent variables (identified success factors) and dependent variable (SPI implementation in GSD) as given below. 4.1.1 Management commitment On the basis of frequency of 'management commitment' (81%), it is the most common success factor for process improvement implementation. According to Sulyman et al. [LT2] management commitment is the extent to which the higher and lower level management in an organisation support, finance, realise and participate in SPI program. Sulyman et al. [LT2] identified management commitment as a positive influential factor for SPI implementation. Ramasubbu [LT5] emphasised on the commitment and participation of organisational management in SPI implementation activities. He mentioned that management should evaluate and track the contribution of GSD team members in process improvement activities. Goldenson and Herbsleb [LT3] conducted a survey in 56 software companies. They argued that some organisations were successful and others were not and have stated that respondents from organisations with effective SPI efforts inclined to agree that managers actively monitor the SPI program and they allocate sufficient resources for SPI implementation. Rainer and Hall [LT4] conducted a study and highlighted that management commitment has been considered as one of the most significant human factors. 4.1.2 Staff involvement The findings of this study also show that 'staff involvement' (63%) is the second most common factor in the extracted list of success factors. It is the degree to which employees of the organisation owns, trust, acquire, perform and believe the SPI program [LT2]. Ramasubbu [LT5] conducted a study in GSD industry and highlighted that without the interest of staff members the implementation of SPI program is impractical. Emam and Koru [LT6] conducted a survey in various organisations and concluded that coordination, commitment and collaboration among SPI team members are important for successful process improvement program. Ramasubbu [LT5] discussed the availability and involvement of SPI team members in GSD. He discussed that the willingness of team members to contribute towards the successful SPI program is an important success factor in a distributed environment. 4.1.3 Allocated resources In the findings of this study, 'allocated resources' (63%) was identified as a CSF. According to Sulyman et al. [LT2], allocated resources are the dedication of time and resources to process improvement implementation. Niazi et al. [LT31] emphasise on the availability of various resources in the form of staff, documentation, time, planning, tools and budget investment. Ullah Khan et al. [LT1] conducted an empirical study in offshore outsourcing organisations and acknowledged that organisations must commit their resources and constantly motivates people to contribute. Niazi et al. [LT16] organisation should brief the managers towards the allocation of all the required resources in order to achieve long-term SPI benefits. 4.1.4 Pilot projects In this study, 'pilot projects' (57%) is an affective and CSF as per results. Pilot projects are the testing phase for all the selected models or standards for SPI implementation. According to Kautz and Nielsen [LT8], after appropriate planning and conducting the pilot implementation experiment of SPI models or standards, the results of pilot implementation should be illustrated to other departments to get support and confidence. Niazi et al. [LT31] described that small-scale process improvement implementation is necessary to see that how effective it is within a specific department. In order to measure the present process improvement readiness and skills, a pilot implementation is vital for experts [LT31]. Niazi et al. [LT11] demonstrate that during pilot implementation, the practitioners can choose the extent of training, resources and commitment, which is needed for implementing the SPI practices across the organisation. 4.1.5 Setting SPI goals 'Setting SPI goals' (48%) is identified as an important success factor among the identified factors. Sulayman et al. [LT2] defined 'setting SPI goals' as the extent to which everyone can make the translation from leading management goals towards the goals that everyone has been asked to accomplish. Setting 'relevant goals' means the efforts for improvement to contribute towards the success [LT2], while setting 'realistic goals' denotes that the goals might be attained in the foreseeable future with a reasonable amount of resources. The purpose of setting SPI goals has been acknowledged as an important factor to accomplish business objectives [LT11]. Petterson et al. [LT12] have developed a light-weight process assessment framework. While formulating the framework, several CSFs were discussed. The important factor associated with the study is given as 'setting SPI goals'. Iversen et al. [LT13] said that since the SPI program seems costly and time consuming, consequently clearly explained goals a
Referência(s)