Open source software development has emerged as one of the most interesting ... primary source for acquiring the list of companies was a national business contact ..... Education & Training, Norfolk, Virginia, USA, IEEE Computer Society. pp.
1(8)
Role of Open Source Software in Commercial Software Development: A Quantitative Analysis Uolevi Nikula, Sami Jantunen Department of Information Technology, Lappeenranta University of Technology, P.O. Box 20, FI-53851 Lappeenranta, Finland {Uolevi.Nikula|Sami.Jantunen }@lut.fi
Abstract Open source software development has emerged as one of the most interesting phenomenon in software engineering today. Unfortunately only little hard data is available to quantify the breadth and depth of the interest in open source software in commercially operating software houses. As a part of a local initiative to align software engineering research and education better with the local industry it was decided to survey the local industry to collect data on the current practices. Due to the globally recognized interest in open source software (OSS) the question arose on the local interest in OSS and some explorative questions on the topic were included in the survey. The distinguishing characteristics of the study were focusing on a geographically bounded area and trying to identify and study alls the software companies in the region. Overall the study sought to increase the general understanding of the current practice in the region to be able to start improving them. Keywords: survey, state of the practice, open source software
1. Introduction Open source software development has emerged as one of the most interesting phenomenon in software engineering today. Unfortunately only little hard data is available to quantify the breadth and depth of the interest in open source software in commercially operating software houses. There are some quantitative studies on open source software development but it seems that so far the interest in these papers has focused on understanding the community of open source developers (e.g. Dempsey et al. 1999; e.g. Ghosh and Prakash 2000). As a part of a local initiative to align software engineering research and education better with the local industry it was decided to survey the local industry to collect data on the current practices. Due to the globally recognized interest in open source software (OSS) the question arose on the local interest in OSS and some explorative questions on the topic were included in the survey. Overall the central tenets of the survey were twofold. First, the study was conducted in a geographically well bounded area – the South Karelia and Kymenlaakso provinces which are collectively called the South-East Finland. Second, the goal was to identify and survey all the companies in the area that practiced serious commercial software development. In this paper the conduct and results of the current state survey are presented.
2(8)
The desire to involve all the software companies in the region in the survey had two reasons. First, a comprehensive analysis on the current practice in all the software companies was expected to effectively remove uncertainty about what the software companies actually do and, thereby, increase understanding of the state of the practice. Second, awareness of the current practices was expected to make it possible to focus attention on why companies operate the way they do and also make it possible to start improving the practices. It has been claimed (March and Smith 1995) that both of these aspects are needed in information technology research, and studies on both of these aspects have been published. For example, IEEE Software magazine had a special issue on the state of practice in software engineering in 2003 (Glass 2003) and sporadic reports on regional process improvement efforts have been reported from different parts of the globe (Habra et al. 1999; ICT West 2004; Jenkins 2004). Regional state of the practice studies appear rear in literature but software development worldwide in general seems like an interesting topic today (Cusumano et al. 2003). However, as the overall goal of the present study was to start improving the software engineering practices in local industry, it was found critical that the improvement actions are based on hard data on the current situation rather than anecdotal evidence and hunches. For example, the collected data on the interest and use of OSS in commercial software houses makes is possible to align the software engineering research and education better with the needs of the local industry. The rest of this paper is structured as follows. Section 2 describes the research method used and Section 3 reports the results of the study. Section 4 provides a discussion on the results and Section 5 closes the paper with conclusions.
2. Research Method The present study focuses on software companies in the South Karelia and Kymenlaakso provinces which are collectively called South-East Finland. The primary source for acquiring the list of companies was a national business contact and marketing information provider (Blue Book 2004) where the companies were identified based on a national adaptation of the European standard industrial classification (The Commission of the European Communities 2001) using Section K 72 Computer and related activities as the selection criteria. The initial list was supplemented with companies found from different sources such as local technology centers, company listings from the provincial federation, listings from previously conducted research efforts, and web searches. The final selection criteria required software development for commercial purposes involving more than one person. The survey was implemented in two parts. First, key informants of the companies were interviewed using a structured questionnaire, and second, a link to a web survey was sent to selected employees of the companies. The employees to answer the web survey were acquired from the key informants at the end of their interviews. The thirteen-page key informant questionnaire form was developed based on a literature survey and numerous other surveys found thereby. In this paper the data from these interviews is mainly used to establish the industrial context of the present study. The four-page employee survey focused on the use of tools, technologies, and processes. Both forms were piloted in another region in six
3(8)
companies with comparable characteristics as the target companies, and changes were made to reduce ambiguity in the forms. The key informant interviews were conducted with the help of a laptop and a data projector so that the interviewee(s) could see the form and how it was filled in by the interviewer. The interview sessions were conducted at the interviewee’s premises when possible, they were audio recorded for future reference, and took from 1 hour and 10 minutes to 3 hours and 40 minutes to complete with an average of 2 hours and 10 minutes. Since three separate people conducted the interviews, the researcher responsible for the questionnaire development attended the first three or four interviews to assure the uniform performance of the interviews. Each employee named for the web survey received a personal link to the survey form by e-mail, so that the responses to the survey could be monitored; if no response was received within a week, a reminder was sent to each respondent. The data from both surveys was stored in a database and later analyzed with the help of a statistical software package (SPSS Inc. 2004) and a spreadsheet program (Microsoft Excel).
3. Results In this section the results of the study are reported. Section 3.1 reports the company and employee characteristics and 3.2 focuses on the attitude towards OSS. 3.1. Company and Employee Characteristics The companies were studied on a business site basis since many of the sites were subsidiary locations for the companies with headquarters located elsewhere in Finland. The conducted site search found 77 sites that were deemed interesting from the point of view of the present study and 62 of them (81 %) agreed to be interviewed. 89 % of the interviewed sites were willing and able to report the share of software related activities of their revenue: 64 % got more than 60 % of the revenue from software business, 18 % got 31 – 60 %, and 18 % got 0 – 30 % of the revenue from software business. The companies were operating mostly with business customers and the largest customer type was large companies which represented 45 % of the total revenue; small and medium sized companies represented 40 %; the public administration represented 15 %; and private consumers represented only 1 % of the total revenue. Considering the customers by their industry sector, the forest industry represented 20 % of the total revenue of the companies, information and communication technology 19 %, metal industry 10 %, energy and environment 10 %, logistics 5 %, and other sectors 35 % of the total revenue. In the other sectors the largest single customer type was the public sector which formed 29 % of the group. The number of people in each site was generally small, and the largest site had 168 employees. From the employee count point of view the largest group of sites had from 5 to 9 employees (35 %) while the most typical size of a whole company was from 10 to 49 employees (33 %) (Table 1). The two biggest sites were part of a corporation that claims to be the largest information technology services provider in the Nordic countries; altogether this corporation employed 468 of the 1196 (39 %) full time employees in the interviewed software houses in the region. From the people working in all the interviewed sites, 63 % were software developers and the rest included management, trainees, and sales and other personnel.
4(8)
Table 1. Number of employees counted per business site and per company Employees 2–4 5–9 10 – 49 50 – 249 Over 249
Per Site Count Percentage 17 22 20 3 0
27 % 35 % 32 % 5% 0%
Per Company Count Percentage 11 18 20 6 6
18 % 30 % 33 % 10 % 10 %
The data from the key informant interviews revealed some interesting aspects of the actual way of working. Observations about the software the companies produced include the following: 89 % of the companies had some kind of commercial off the shelf software products available; 70 % of the bespoke projects and products were completed with less than five person years of effort; and in general the companies operated with non-critical applications. Namely, the companies reported that a failure in the end products could result in a loss of human lives in only 5 % of the companies, halt the normal company operation in 53 %, result in significant economical losses in 50 %, cause distraction to normal operation in 92 %, and result in general irritation and dissatisfaction in 97 % of the companies. In general, 82 % of the companies had carried out some process improvement activities during the past 12 months but only 29 % of them had utilized the established quality system standards (ISO-9001) or maturity models (CMM, CMMI, SPICE) in these activities. Two companies themselves noted that they had an ISO certificate and one was applying for it whereas numerous companies had utilized the standard in their development work. The key informant interview was closed with a request to name from 1 to 3 typical project managers and from 2 to 6 typical developers with different levels of experience for the employee survey. The number of names received varied from zero to nine with a total of 165 named respondents; from these 128 (78 %) from 54 companies completed the web survey. 83 % of the respondents worked as developers, 30 % as managers, and 13 % played both of these roles. From the respondents (n=118) 26 % had less than 5 years of experience in software engineering, 48 % had from 5 to 10 years, and 25 % had over 10 years of experience in software engineering. The experience in the present company (n=124) was less than 5 years for 57 % of the respondents; 31 % had from 5 to 10 years, and 11 % had over 10 years of experience in the present company. 59 % of the respondents had an education in a field related to software engineering, 37 % had other technical education, 10 % had business education, and 10 % had some other education. Considering the level of education, a Master’s degree or an equivalent was held by 30 % of the respondents, a Bachelor’s degree by 29 %, a vocational degree by 20 %, and a degree higher than a Master’s degree was held by 1 % of the respondents; 9 % had some other degree and 16 % had no degree at all. The areas of expertise of the respondents varied as follows: 66 % were specialized in design, 60 % in the application domain, 44 % in databases, 38 % in project management, 36 % in requirements, 40 % in architecture, 36 % in testing, 34 % in maintenance, 30 % in process improvement, 29 % in customer support, 25 % in usability, and 22 % in quality.
5(8)
3.2. Attitude towards Open Source Software The key informant survey studied the interest in the OSS with the question “Do you utilize open source software in your software development?” This question had five non-exclusive response options: 1. No we do not 2. We are investigating the possibilities to utilize open source software in our software development 3. We utilize open source software in our internal activities 4. We utilize open source software in our commercial activities 5. Our commercial software development is based on open source software. 35 % of the companies (n=62) reported that they do not utilize OSS in their business (Figure 1). 48 % (30) of the companies reported that they are studying the possibilities to utilize OSS in their business and at the moment four of these companies did not utilize OSS in any way. 50 % (30) of the companies reported that they utilized OSS internally and 18 of these was studying possibilities for wider utilization. 44 % (27) reported that they utilized OSS it in their business; of this group 23 companies utilized OSS internally and 16 were studying possibilities for wider utilization. Only 5 % (3) of the companies stated that their business was based on OSS. Looking at the selected combinations of these options only two unexpected combinations were found. First, three companies utilized OSS in their commercial activities but did not do so internally; further these companies studied possibilities to larger utilization of OSS. Second, one of the companies that based its business on OSS and used it in internal operations, but did not use it in commercial activities or study possibilities for wider utilization.
60 % 48 %
50 % 40 %
50 % 44 %
35 %
30 % 20 % 10 %
5%
0%
Not utilized
Possibilities are studied
Used internally
Used in business
Basis of the business
Figure 1. Interest in OSS, n=62 The web survey included two questions on the operating systems and database systems used. In the operating systems question four options were given in addition to the Not used option which were Microsoft operating systems, Unix based systems, embedded systems, and other. This question was answered by 128 people from 54 companies and looking at the operating systems used (Figure 2 (a)) the
6(8)
Microsof t
81 %
Unix
Other
Object based
9% 20 %
24 % 4%
Other
2%
0%
30 %
Open source
20 %
Not used
59 %
File based
35 %
Embedded
Commercial
2%
Not used
40 %
60 %
80 % 100 %
0%
23 % 20 % 40 % 60 % 80 % 100 %
(a) Operating Systems (b) Databases Figure 2. Operating systems and databases, n=128 dominance of Microsoft operating systems was clear (81 %) (e.g., Windows 2000, XP). However, Unix based systems (35 %) (e.g., UNIX, Linux) and embedded systems (20 %) (e.g., Symbian, PalmOS) were also well represented in the participating companies. In the same vein databases (Figure 2 (b)) were clearly dominated by commercial solutions (59 %) (e.g., Oracle, MS-SQL Server, MSAccess), and file based solutions (30 %) (e.g. XML) were more common than OSS databases (24 %) (e.g., MySQL, PostgreSQL).
4. Discussion The collected statistical data bears little surprises in it and mainly confirms the general perceptions about interest in OSS. The fact that 5 % of the companies based their business on OSS, 35 % did not utilize OSS at all, and 44 – 50 % of the companies were doing something between these two extremes suggests that OSS is in its early adoption phases (Rogers 1995). Adopting a new business model can be expected to differ quite a bit from adopting a technological innovation like a mobile phone or a method (Riemenschneider et al. 2002) but since basic ideas on setting up a business based on OSS have been published (Hecker 1999), the ground for innovation research in this area may be paved. In the key informant interviews some OSS related issues were discussed. In one company the managing director told that he was interested in OSS database information to consider it as an option for commercial databases but he could not find reasonably priced training from anywhere in Finland. Consequently he said that they were forced to stick with the commercial databases for the time being. In another company the managing director said that they had been looking for business opportunities with OSS couple of years ago but failed to find anything solid enough for business activities. Consequently they decided to become Microsoft partners, and he claimed this decision and partnership very positive ones for their company. In the course of the conducted interviews two further companies brought up the Microsoft Certified Professional training programs and the benefits they brought to the companies. This paper has some obvious limitations. It was not possible to collect accurate data of the site revenues and no provision is made to the companies’ economical importance in the region, but results are reported on the basis of site counts. The employee survey sampling does not reflect the size of the companies and their selection was done by the key informants who could distort the respondents’ profiles; in some cases the key informants themselves responded to the employee survey since they also participated in daily software development work. The
7(8)
applicability of the results in any other region is questionable, and finally, the study provides only a snapshot of the current practices and no data was collected on future plans or needs.
5. Conclusions In this paper results of a survey on the role of open source software in commercial software development in South-East Finland were reported. The survey was conducted in a regionally bound area and managed to engage 81 % of the local software companies in the study. Overall the study aimed at demystifying the current practices so that research and education on software engineering could be aligned with the real needs of the local industry. The results of the study confirmed the perceptions about the general attitude towards open source software but revealed some aspects that had not been thought about before and, most importantly, the collected data provided hard data on the local situation. The conducted interviews brought up the fact that individual companies may have well argued reasons to base their business solely on commercial software or open source software, but the conducted survey provides evidence on the increasing importance of open source software. Overall the collected data suggests that educational and research institutions should be prepared to include also open source software development related topics in their knowledge areas. The future research should include continued interaction with the industry to confirm that the results of the study reflect their feelings about the current practice, to explore the reasons for the current situation, and to find out more about the perceived future needs. On the academic side it is evident that there is a clear need to increase the understanding of and skills in open source software area to be prepared to serve the industry better. This includes also helping the students to build a solid understanding of open source software and its realities in commercial software industry.
Acknowledgements This study was conducted as a co-operative effort between Lappeenranta University of Technology, South-Carelia Polytechnic, and Kymenlaakso Polytechnic and was funded by the National Technology Agency and the Employment and Economic Center of Finland; more information about the project can be found at http://www.it.lut.fi/katapultti.
References Blue Book (2004). Bluebook, Nordic TDC Directories. http://www.bluebook.fi. Accessed 27 September 2004. Cusumano, M. A., A. MacCormack, C. F. Kemerer, and B. Crandall (2003). "Software Development Worldwide: The State of the Practice." IEEE Software 20(6): 28-34. Dempsey, B. J., D. Weiss, P. Jones, and J. Greenberg (1999). A Quantitative Profile of a Community of Open Source Linux Developers. SILS Technical Report TR-1999-05. Chapel Hill, North Carolina, USA, University of North
8(8)
Carolina at Chapel Hill: 18 p. Available at http://ils.unc.edu/ils/research/reports/TR-1999-05.pdf. Ghosh, R. A. and V. V. Prakash (2000). "The Orbiten Free Software Survey." First Monday 5(7). http://firstmonday.org/issues/issue5_7/ghosh/index.html. Glass, R. L. (2003). "The State of the Practice of Software Engineering." IEEE Software 20(6): 20-21. Habra, N., E. Niyitugabira, A.-C. Lamblin, and A. Renault (1999). "Software Process Improvement in Small Organizations Using Gradual Evaluation Schema." International Conference on Product Focused Software Process Improvement, Oulu, Finland, Technical Research Center Finland. pp. 381396. Hecker, F. (1999). "Setting Up Shop: The Business of Open-Source Software." IEEE Software 16(1): 45-51. ICT West (2004). ICT West, Tampereen teknillinen yliopisto, Porin yksikkö. http://www.tut.fi/public/index.cfm?MainSel=7476&Sel=7479&Show=7874 &Siteid=109. Accessed 24 November 2004. Jenkins, M. (2004). "PRO-SOFTWARE: A Government-Industry-Academia Partnership that Worked." 17th Conference on Software Engineering Education & Training, Norfolk, Virginia, USA, IEEE Computer Society. pp. 92-97. March, S. T. and G. F. Smith (1995). "Design and Natural Science Research on Information Technology." Decision Support Systems 15(4): 251-266. Riemenschneider, C. K., B. C. Hardgrave, and F. D. Davis (2002). "Explaining Software Developer Acceptance of Methodologies: A Comparison of Five Theoretical Models." IEEE Transactions on Software Engineering 28(12): 1135-1145. Rogers, E. M. (1995). Diffusion of Innovations. 4th edition. New York, The Free Press. SPSS Inc. (2004). SPSS Statistical Analysis Software. http://www.spss.com/. Accessed 28 September 2004. The Commission of the European Communities (2001). NACE Rev. 1.1: Statistical Classification of Economic Activities in the European Community, The Commission of the European Communities,. http://europa.eu.int/comm/eurostat/ramon/nace_rev1_1/nace_rev1_1_en.ht ml. Accessed 27 September 2004.
