Software Process Improvement in Small Latin-American Organizations:
COMPETISOFT Project
Hanna Oktaba°, Mario Piattini*, Félix García*, Francisco J. Pino+, Claudia Alquicira°, Francisco Ruiz*, Tomás Martínez*
° Facultad de Ciencias
National Autonomous University of Mexico (UNAM) Ciudad Universitaria, 04510 México D.F., Mexico [email protected], [email protected]
* ALARCOS Research Group
Information Systems and Technologies Department University of Castilla-La Mancha
Paseo de la Universidad, 4 – 13071 Ciudad Real, Spain
{Mario.Piattini, Félix.García, Francisco.RuizG, Tomas.Martinez}@uclm.es
+ IDIS Research Group
Electronic and Telecommunications Engineering Faculty University of the Cauca
Calle 5 # 4 – 70, Popayán, Colombia.
Software Process Improvement in Small Latin-American Organizations:
COMPETISOFT Project
ABSTRACT
This chapter deals with the creation of the software reference process, assessment and improvement models adapted to the characteristics of the software industry in Latin America, mainly composed of small and medium-sized enterprises (SMEs), and presents the
COMPETISOFT project, whose main aim is to provide the software industry in Latin America with a reference framework for software process improvement and certification, which will enable them to be more competitive in the global market. The COMPETISOFT approach is based on solutions which have previously shown their practical usefulness, especially the MoProSoft process model. While relying on feedback from the MoProSoft experience, along with its process reference and evaluation models, COMPETISOFT enhances and refines them in the light of experience and lessons learned.
Keywords – Process Improvement, IS Maturity, Software Evaluation, Software Quality, IS Models, SMEs, Management. Software process models..
INTRODUCTION
The software industry is a highly important economic entity in every country in the world, and constitutes an especially important opportunity for developing countries, as is the case of many of those in Latin America, which mainly comprise small and medium-sized enterprises (SMEs). Indeed, SMEs account for about 90 percent of Latin America’s formal businesses and generate between 40 to 50 per cent of all employment. Such companies face serious problems as they reach maturity however, as in many cases there is no common software development process known to a given firm, which leads to chaotic performance affecting the whole organization (Batista et al., 2000) and, obviously, its products. This situation is
especially critical in Latin America’s incipient software industry, with its problems of lack of competitiveness and consequent limitations of growth (Mayer&Bunge, 2004).
For this reason, in recent years Latin American SMEs have been attempting to improve the capability of their software processes, as a fundamental step towards increasing product quality, and they address two main concerns: their image, a key factor for export purposes and establishing and maintaining a position in the global marketplace; and the efficiency and effectiveness of software process management.
Many of these businesses have the goal of process improvement through the deployment of the reference models proposed by the Software Engineering Institute – SEI – (CMMI®) or the International Organization for Standardization – ISO – (9001:2000, 12207, 15504), but, as numerous studies have claimed, assessments based on these models are expensive and time- consuming, difficult for small companies to perform, their process model structure is too complex and the return on investment is late in appearing (Hareton et al., 2001) (Maller et al., 2005) (Saiedian et al., 1997).
In Latin American SMEs the situation is especially dramatic, owing to the absence of well- defined software life cycles, and also the important cultural problem occasioned by importing
and adopting models defined in other countries without suitable adaptation. Indeed (Zahran, 1998) says that if the process does not match the culture of the organization, it will be rejected by its body like a mismatched transplanted organ. A similar problem is set out in (Dyba, 2005), where the main cultural differences in the success of software process improvement between Europe and the USA are highlighted. The SEI in (Chrissis et al., 2004) even insists on how expensive and difficult it is to implement CMMI and SCAMPI. If this is the situation in US companies, then the cost of applying these models will be prohibitive to those in Latin America, to say nothing of considerations of copyright and certification payments to the USA.
Owing to limited resources, small companies need external assistance in planning and implementing process improvement to keep abreast of state-of-the-art Software Engineering research and practice. Furthermore, the current International Life Cycle standards (ISO/IEC 12207 and ISO/IEC 15288) do not explicitly address the needs of SMEs. The new ISO/IEC JTC1 SC7 Working Group 24, which was created to develop the “Software Life Cycle Profiles and Guidelines for use in Very Small Enterprises (VSE)”, recognizes the need to adapt ISO 12207 increasingly to small settings.
A number of proposals, mainly from the ISO and SEI, exist in the literature seeking to deal with these issues and thus adapt process assessment and improvement models to the special characteristics of SMEs (see next section). Especially worthy of attention are the recently published proceedings of the “First International Workshop for Process Improvement in Small Settings” (Garcia et al., 2006) in which the challenges of process improvement in SMEs are discussed.
Therefore, countries such as Mexico have developed local programmes to promote the improvement of their software industry, and this has led to the development of the MoProSoft process reference model (Modelo de Procesos para la Industria de Software, Process Model for the Software Industry)(Oktaba, 2005). On the basis of the MoProSoft experience, and in collaboration with several experts, a new initiative has been promoted: the COMPETISOFT project, which involves eleven Latin-American countries plus Spain and Portugal. Its main aim is to provide Latin-American countries with a reference framework for the improvement and certification of their software processes which will help them to be more competitive in the global market.
In this chapter, the current convergence of efforts to improve the software industry in Latin American countries is reported in the context of the COMPETISOFT project. First we summarize the related works, and then the COMPETISOFT project is presented, with a description of its scope and characteristics. Finally, conclusions are drawn and future work is described.
RELATED WORKS
In literature, diverse proposals and initiatives are to be found which address the tailoring of assessment and improvement initiatives for the special characteristics and problems of small and medium-sized enterprises.
Some methods adapted to ISO 15504 for SMEs are SPINI (An approach for SPI Initiation) (Mäkinen et al., 2000), TOPS (Toward Organised Processes in SMEs) (Bucci et al., 2001) and the RAPID (Rapid Assessment for Process Improvement for Software Developed) (Cater- Steel et al., 2005), which is based on ISO/IEC 15504:1998. In this last proposal, developed by
the Australian Software Quality Institute, the scope of the assessment is limited to eight key processes (requirements elicitation, software development, configuration management, quality assurance, problem resolution, project management, risk management, and process
establishment) and the assessment model only includes questions relating to levels 1 to 3. The model was applied to 22 small Australian software development firms, 15 of which reported some improvements in their processes.
With regard to SEI assessment and improvement models, some representative tailoring proposals are reported in different studies, such as (Batista et al., 2000), (Blowers et al., 2006), (Coleman et al., 2005), and (Serrano et al., 2006), among others.
With regard to Latin American initiatives, the following proposals deserve special attention:
The MARES method (Método de Avaliação de Processo de Software) (Anacleto et al., 2004) has been developed in Brazil. It is based on capability dimension from ISO/IEC 15504:2003 Part 5 – An exemplar Process Assessment Model –. MARES, a set of guidelines for conducting 15504- conformant software process assessments, focuses on small companies. MARES is not yet another method but rather a set of well-structured guidelines for conducting 15504-conformant software process assessments in small
companies. MARES provides guidance for identifying target process profiles and selecting high-priority processes to assess an organization on the basis of its business goals and model (concerning its product, revenue, services, and implementation strategies) as well as its process maturity and growth stage. It also adapts a growth pattern model to represent a small company’s evolution from its existence (focusing on it becoming a viable business) to its survival (negotiating between expenses and sales income) to its success (growing in size and profitability).
The mps Br project (Weber et al., 2005), another Brazilian initiative, is based on ISO/IEC 12207:2002, CMMI and ISO/IEC 15504:2003, and comprises two models for the software improvement process: a Reference Model (MR mps) and a Business Model (MN mps). The maturity level is organized in two dimensions: capability and process, with process
maturity divided into seven levels: Optimized, Managed Quantitatively, Defined, Almost Completely Defined, Partially Defined, Managed and Partially Managed. There are twenty- one process areas, which are attributed to each maturity level based on the levels of CMMI.
The aim is to ensure a gradual and fully appropriate implementation of the model in Brazilian SMEs.
The SIMEP-SW (“An Integrated System for the Improvement of Software Development Processes”) is a proposal developed in Colombia. It integrates elements from improvement to capability, process and assessment models which are internationally recognized but tailored to the specific characteristics of Colombian SMEs. The main result of the SIMEP- SW Project is Agile SPI (Software Process Agile Improvement) (Hurtado et al., 2006), whose main components are: (i) Agile Process: An agile process which guides the process improvement programme; (ii) Light Quality Model: A lightweight capability and
assessment model of productive process; (iii) Framework PDS: a conceptual and technical element to support processes and (iv) Light Metrics Quality Model: A lightweight model of measurements for the productive process.
The Mexican Government promoted the MoProSoft project (Oktaba, 2005) together with a corresponding method for process assessment – EvalProSoft–(Oktaba, 2004), with the aim
of providing the Mexican software industry with a model based on the best international practices whilst being easy to understand, simple to apply and economical to adopt. It sought to assist organizations in standardizing their practices in the assessment of their effectiveness and in the integration of ongoing improvement.This model builds on the well-known practices of SW-CMM, ISO 9000:2000, PMBOK and others and offers a new process structure, some new process documentation elements, a more precise process relationship and an explicit process improvement mechanism. Trials of MoProSoft and EvalProSoft in four Mexican companies confirmed the suitability of the model for SMEs with low maturity levels, borne out by the improvements achieved and the low cost of process adoption. In August 2005, MoProSoft was approved as Mexican standard NMX- 059-NYCE-2005. In a Bangkok meeting in May 2006, the ISO/IEC JTC1 SC7 Working Group 24 decided to use the Mexican standard as a basis for the development of profiles and guidelines for Very Small Enterprises (VSE).
THE COMPETISOFT APPROACH
In 2005, several researchers and practitioners from different Latin-American countries recognized the importance of an improvement and certification framework for SMEs, and under the direction of Professors Hanna Oktaba and Mario Piattini suggested the
COMPETISOFT project to CYTED (Programa Iberoamericano de Ciencia y Tecnología para el Desarrollo – Ibero-American Science and Technology Development Programme–), which was created in 1984 for multilateral scientific and technological cooperation and is supported by 21 Latin-American countries plus Spain and Portugal. The CYTED program has the objective of contributing to the harmonious development of the Ibero-American region by establishing cooperation mechanisms between university research groups, R&D institutes, and innovative companies in the countries involved, with a view to obtaining scientific and technological results transferable to productive systems and social politics.
The research method applied in the definition, refinement and application of the
COMPETISOFT model is Action-Research (A-R), which is a collaborative research method merging theory and practice. A-R is focused on building new knowledge through the
identification of solutions to real problems, which is achieved by the intervention of researchers in the real world of practitioners and by the results of this experience being equally beneficial to both groups. The application of A-R is based on continual feedback between the researchers and the companies involved. Figure 1 provides a summary of the A-R application.
Research Groups participating in the project (researchers)
Software Process Improvement of the Latin American SMEs
(object of research) Latin American SMEs
(stakeholders)
companies and organizations participating in the project
(critical reference group) Proposals
Refined results Research
results
Application results
Deliverables (norms, models, methods, tools, handbooks, etc.)
Figure 1. Action-Research application to the COMPETISOFT project
The participants in the COMPETISOFT project could be grouped into two main categories:
Researchers, from the following universities: National Autonomous University of Mexico, Mexico; New University of Lisbon, Portugal; University of São Paulo, Brazil;
University of the Andes, Venezuela; Federico Santamaría University, Chile; National University of Comahue, Argentina; University of Castilla-La Mancha, Spain; International SEK-ICAPI University, Ecuador; University of the Cauca and EAFIT Colombia;
University of Information Technology, Cuba; Institute of Technology, Costa Rica; National University of La Matanza, Argentina; Catholic University of the Maule, Chile; University of the Republic, Uruguay and Pontificial Catholic University of Peru.
Critical Reference Group, a representative group of the problem to be solved which participates in the research process, although less actively than the researchers themselves.
In the COMPETISOFT project this reference group is composed of three main types of organizations: a national standards body, the Argentinian IRAM (Argentinian Institute for the Standardization and Certification); a government body, the Government of the Neuquén region in Argentina and SMEs such as the Spanish firms Enxenio and Technical Systems of the Spanish Lottery, Ultrasist of Mexico, MV Systems of Ecuador, and Parquesoft Popayán of Colombia.
To develop the COMPETISOFT project we studied different Latin-American initiatives such as: MoProSoft, MPS.BR and Agile SPI. The Spanish methodology METRICA v3, which is promoted by the Spanish Ministry of Public Administration, was also considered, as it too seeks to achieve the improvement of software processes and products. A general overview of COMPETISOFT is given in Figure 2:
Figure 2 – Overview of the COMPETISOFT Project
As Figure 2 shows, we developed COMPETISOFT by borrowing heavily from well-known process reference models intended for small companies, especially MoProSoft. In fact, COMPETISOFT can be conceived as an evolution of the MoProSoft model with all the experience in software process development and improvement gained by researchers and practitioners during the COMPETISOFT project. The main objective is to develop the new Process Reference and Evaluation Model, which enhances MoProSoft and EvalProSoft, and a new Process Improvement Model based on Agile SPI.
The COMPETISOFT Process Reference Model
The COMPETISOFT process reference model is based on MoProSoft. The processes are organized in three categories: Top Management, Management and Operation. Its purpose is to provide specialized processes for each functional group of the software development
enterprise. The following figure presents the MoProSoft category and process structure in the form of a UML package diagram.
Figure 3. MoProSoft Process categories and process names.
The proposed COMPETISOFT Process Reference Model improvements and refinements are the following:
Business Management Process - we need to include virtual enterprise management and inter-company connectivity. Nowadays these are increasingly important topics for companies participating in clusters or virtual nets, and key requirements to guarantee the survival of SMEs in today’s marketplace.
Process Management Process - we need to improve quality assurance and internal and external assessments. We have developed a self-assessment questionnaire that could help SMEs in their first contact with the assessment and improvement of their process maturity.
Project Management Process - we have to include a set of measures and indicators for different maturity levels. Another critical point to be tackled here is the improvement of estimation techniques, a fundamental need of SMEs but one that is difficult to understand and apply in these settings.
Resource Management Processes - we need to emphasize the importance of reusability by means of the development of a knowledge base to be populated with experiences structured according to the processes in the process reference model. To this end, other similar experiences, as reported in (Kurniawati et al., 2006), will be considered. Indeed,
COMPETISOFT gives great importance to the experience base from the outset and at all organizational levels, regardless of the quality of the components stored in that base, as they may all be useful. We also recognize the value of a more formal yet still lightweight method of experience elicitation suited to the use of a small organization, providing guidance and structure to assist users in creating more experiences for the base. Other important issues to address are documentation and configuration management.
Development and Maintenance Process - we consider it very important to separate this process into two parts. The nature and characteristics of development and maintenance are very different, and many of the techniques, tools, methodologies, etc., of development are not directly applicable to maintenance. Indeed, many SMEs have to develop pure software maintenance projects, so it is very important for them to apply specific maintenance methodologies. In this respect, the COMPETISOFT approach will adapt the MANTEMA maintenance methodology (Polo et al., 2002) to the special characteristics of SMEs.
Besides the improvements proposed for specific processes, two aspects common to all processes will be the incorporation of free and open-source tools, a key element for SMEs to reduce costs, and the development of specific techniques for the improvement of software usability.
The COMPETISOFT Evaluation Model
The COMPETISOFT evaluation model is based on the EvalProSoft, for which the following improvements and refinements have been proposed:
Measures – we need to define a set of measures to evaluate the performance and capability of software processes. The aim is to help SMEs carry out their own internal assessments by reducing subjectivity and making the process more formal. The measures are grouped into two main types (see figure 4):
The “capability measure” to evaluate process capability (from level 1 to 5) on the basis of the process attribute’s indicators (PAIs) from ISO/IEC 15504-5:2006 (E).
The “performance measure”, based on some of the elements defined in MoProSoft, to evaluate process performance.
Instruments – we need to build the collection of instruments to support the measurement process related to the measures proposed. There must be an instrument for each attribute of the capability dimension and for all the MoProSoft processes in the process dimension.
Tools – we need to develop a software tool to support the application of process performance and capability evaluation instruments.
Figure 4 –COMPETISOFT Evaluation Model The COMPETISOFT Improvement Model
The COMPETISOFT improvement model is based on Agile SPI (Hurtado et al., 2006). Its purpose is to provide a guide for the carrying out a programme of software process
improvement in small and medium companies. The aim is to provide SMEs with a software improvement framework that uses the process reference and evaluation models previously described and furthermore, to provide improvement infrastructure, the techniques and tools to support improvement programs.
This model integrates some basics of the agile paradigm and the IDEAL® (McFeeley, 1996) model has also been considered for its development. The improvement process life cycle is highly influenced by the iterative and incremental approaches of many methodologies such as RUP, XP, Scrum, etc. As improvement projects usually cover many requirements and affect the whole organization, some characteristics of these proposals have been adapted to build a complete improvement model which is agile, less bureaucratic, and is sensitive to
management activities (meetings, documentation, infrastructure, etc.).
The model defines an iterative and incremental process consisting of five phases: (i) installation, (ii) diagnosis, (iii) formulation, (iv) improvement and (v) programme review.
Furthermore, it includes a set of disciplines which can be applied in the different phases: (i) training, (ii) SPI management, (iii) evaluation, (iv) result analysis, (v) design, (vi)
implementation, (vii) process configuration management and (viii) learning. By discovering the disciplines which can be used in each phase, the people involved in SPI will always know how to proceed. Moreover, the model is based on improvement cases as it allows companies to create mini-improvement programmes in the context of an overall improvement
programme in order to obtain fast results. Improvement cases are atomic improvement units of the processes selected to be improved. Fast results will allow the visualization of
improvements from the early phases in the improvement project, with the consequent motivation of the SPI staff. This mitigates possible risks from the beginning and allows personnel to focus their efforts more on the most important business areas for the company.
This model is currently being applied in some small software development companies in Colombia. The feedback and experience obtained by researchers and practitioners with the application of improvement programmes during the COMPETISOFT project, will be used to refine and improve the SPI model.
CONCLUDING REMARKS AND FUTURE WORK
Software process improvement and assessment tailored to the special characteristics of Latin American SMEs are two key challenges to address, as companies of this kind need to survive in a more and more competitive global market but they do not have enough money or
resources to apply “heavyweight” approaches. In order to address such challenges the COMPETISOFT project has been set up, with the active involvement and participation of Latin American, Spanish and Portuguese software professionals and academics. Its main results are expected to be:
The development of a common methodological framework suitable for the socio-economic reality of Latin-American SMEs and oriented towards continuous software process
improvement.
The introduction of the process improvement culture in the Latin American software industry and more specifically the training of both academia (researchers and/or lecturers) (“training of trainers”) and a significant number of practitioners from software companies.
Promotion, in the standardization and certification organizations, of the COMPETISOFT methodological framework and the results of its application in order to establish the basis for a commonly recognized mechanism for assessment and certification in the Latin American software industry.
It is important to highlight that the proposed framework, developed by several countries, may have a higher impact than those developed individually, and this may contribute to the
regional integration of the software industry. However, we hope it will lead to the creation of a Latin American Software Engineering community with a good scientific level.
ACKNOWLEDGMENTS
This work has been developed inside the project: “Process Improvement for Promoting Iberoamerican Software Small and Medium Enterprises Competitiveness – COMPETISOFT”
(506AC287) financed by CYTED (Programa Iberoamericano de Ciencia y Tecnología para el Desarrollo).
REFERENCES
