migrating public administrations to open source software - CiteSeerX

MIGRATING PUBLIC ADMINISTRATIONS TO OPEN SOURCE SOFTWARE Paolo Zuliani and Giancarlo Succi Faculty of Computer Science, Free University of Bolzano-Bozen piazza Domenicani, 3 – 39100 Bolzano-Bozen, Italy {paolo.zuliani,giancarlo.succi}@unibz.it *

ABSTRACT In this paper we present some experiences of Public Administrations migrating to Open Source Software and we discuss a project which addresses the migration issue at large. In particular, we discuss COSPA, a research project aimed at studying and supporting the introduction of open source software in the Public Administration. COSPA focuses on office automation and desktop system software and aims at rigorously measuring the effort, costs and benefits of a transition to Open Source. The results of the study would help Public Administration’s managers in their strategic decisions for implementing eGovernment services. The project has also established an “observer” status, by means of which interested parties may access COSPA’s results, thereby fostering collaboration and increasing dissemination of knowledge. KEYWORDS Open source, government, migration, TCO, ROI.

1. INTRODUCTION Open Source Software (OSS) has recently reached a non-marginal level of diffusion. Linux and the Apache web server are found in respectively 30% and 66% of the Internet’s public servers, according to Netcraft’s survey (Netcraft, 2004). It is also well-known that half of the world’s emails are routed by OSS such as Sendmail and Postfix. We thus have some evidence that OSS can work well, at least for the server side of a client-server architecture. On the other hand, the recent FLOSS project (FLOSS, 2002) of the European Union showed that OSS is seldom used for desktop tasks. In particular, OSS for desktop applications (e.g. client operating systems, office automation, etc.) is employed only by 5% of European public institutions and businesses. If we further restrict to the use of OS office automation software that percentage drops to a mere 2%. The difference with system software is indeed striking and it has attracted the interest of researchers. Some authors suggest that the difference is due to the fact that developers of OS system software “knew what they were doing and how to do it” (Fuggetta, 2003). In the case of Linux, the group of developers headed by Torvalds had a clear idea of the requirements of the system being developed, plus they all shared knowledge of the Unix structure (Raymond, 2001). Another reason for the difference in popularity between desktop and system OSS is the fact that the former is actually much younger, and therefore it has not benefited from the same amount of user testing and feedback as OS system software has had. However, as we report in the next section, there is some qualitative evidence that OSS can be successfully employed even for desktop tasks. In particular we briefly report about a transition to desktop OSS in the Municipalities of the Province of Bolzano-Bozen, Italy. The transition focused mainly on the OpenOffice desktop suite. The experience indicated that tackling correctly personnel resistance to change may be the most important factor for a successful transition. Effective training and support are other important factors, while technical or functional problems seem to be quite marginal. That experience served as qualitative basis for a follow-up project at a higher level, the COSPA project. An hypothetical deployment of OSS for desktop applications in a corporate environment, such as Public Administrations (PA), might be hampered by:

• • •

cost of transition from previous solutions (data migration, hardware migration, etc.); interoperability and integration with existing solutions (compatibility of standards, etc.); cost of training personnel for the new tools and hostility to change (fear of the unknown, CV dilution effect, etc.); • reduced productivity of the personnel (caused by the new tools). COSPA aims at studying those issues under a quantitative perspective, in order to provide real numbers of the effort, cost and benefits of a transition to OSS. As far as independent studies on OSS are concerned, we observe a substantial lack. The main effort is the FLOSS project in the 5th EU Framework Programme, which only painted a state of the art of OSS use in Europe.

2. AN EXPERIMENT OF MIGRATION TO OSS We report on the transition to Open Source Software in several municipalities of the Province of BolzanoBozen, Italy. In particular, the transition focused on OpenOffice (OpenOffice.org), a suite of key desktop applications which includes a word processor, a spreadsheet, a presentation manager and a drawing program. The migration experiment was performed by the Consortium of the Municipalities of the Province of Bolzano-Bozen. The transition started in January 2003 and so far it has involved sixty municipalities and about 1,800 desktop PCs. The latter is the number of PCs where OpenOffice (OO) has been deployed, leaving intact the existing user environment (e.g. MS Windows, MS Office, etc). The transition follows a three-step method: 1. visiting the “candidate” site for the transition; 2. hardware/software set up (OO deployment, etc); 3. personnel training. The visit to the site subject of the transition aims at introducing personnel to OO and its advantages/disadvantages over MS Office. Furthermore, a work plan for the deployment of the software is established. OO is deployed using a customised package containing components which are not found in the download version, for example language dictionaries1. The package installs the software in predefined pathnames and also uninstalls any previous version of OO, in order to have the same “disk image” across systems. That also allows a completely remote update of the software. As mentioned, existing software (MS Office in particular) is not removed until expiration of the license. Next, the most used documents are converted on-site to OO format. Finally, municipalities’ personnel are trained on OO. Training was performed off-site and a total of sixty-seven courses were given in 2003. Courses’ length ranged from half-day introductory to two-day intensive sessions. Training was delivered in collaboration with local IT firms. The lessons learnt from the experience may be summarised as follows: 1. The biggest problem appears to be personnel’s resistance to change: the introduction of the new technology it often seen a burden in addition to every day’s work. Personnel have to be persuaded that the new tools, after some training, can provide features that allow faster and easier work. 2. The “brute force” method does not work: i.e. installing OO and wiping the previous tools out (e.g. MS Office) will only result in a even stronger resistance to change. Personnel need time to adapt to the new software. 3. Personnel should be able to produce right after the training: it is crucial to complete set up and test of the work stations before training begins. A delay between training and productive work with the new tools will diminish the effect of the training itself. 4. Converting MS Office macros is in general very difficult. This is perhaps the biggest technical problem encountered, also due to the scarcity of documentation on OO macro development. 5. It is very important for personnel to have a reference point for any problem connected to the transition. This could well be a “champion” inside the same institution, i.e. an employee who is enthusiastic of the new tools and might serve as a guide to other employees. 6. The features offered by OO seemed to be more than adequate for the every day’s needs of the municipalities. 1

OpenOffice 1.1.1, released in March 2004, features an automated tool for downloading and installing any dictionary.

The lessons learnt agree with other studies, such as the IDA Open Source Migration Guidelines (MIGOSS, 2003). In particular, all lessons but 3 and 6 are in line with IDA’s recommendations; the issue of training is not fully addressed in the IDA work, so there is no corresponding party for lesson 3. With respect to lesson 6, we note that most municipalities of our experience have quite small population (a few thousands people) so it might be that the scale of problems addressed is feasible for OpenOffice’s features. Overall, this experiment has qualitatively proved that a transition to OS desktop software is indeed possible and may be successful. However, we have no measures of the effort, cost and benefits of such a transition. A quantitative characterization of the process would be helpful for Public Administrations’ managers in order to better decide on the use of OSS and better allocate resources for an eventual migration. The issue is addressed at European level in the COSPA project.

3. THE COSPA PROJECT The Project aims at introducing, analysing, and supporting the use of Open Data Standards (ODS) and Open Source (OS) software for personal productivity and document management in European Public Administrations (PA). The Project will analyse and support the introduction of ODS and OS solutions in the PA by: 1. Deploying ODS and OS software solutions in several European PAs, and benchmarking their effectiveness through a cost/benefit analysis; 2. Building a European, multilingual, freely-accessible knowledge and experience base by comparing and pooling knowledge; 3. Disseminating the results and the experiences of the study through a series of workshops at regional and European level. 4. In particular, the Project focuses on software for office automation and personal productivity, mainly OpenOffice and Linux for desktops. COSPA is funded with 2,6 million euro by the 6th Framework Programme (FP6) of the European Union and will run from January 2004 to December 2005. The Consortium is made of fifteen European universities and Public Administrations from Belgium, Denmark, Hungary, Ireland, Italy and UK. The structure of the Consortium is centered on university-PA couples. In fact, every PA is co-located with an academic partner, in order to constantly follow the evolution of the transition to OSS.

3.1 Work plan The work plan of the project can be divided in five main activities: 1. gathering and analysis of user requirements from the partner PAs, in order to identify and combine OS software and ODS which fulfill the PA requirements; 2. pilot projects for deploying in the partner PAs the OS desktop solutions developed on the basis of the previous requirement study, in order to enable the subsequent cost/benefit analysis. Deployment will follow a two-step strategy: in the first step the focus is on desktop applications only (mainly OpenOffice). In the second step desktop operating systems (Linux) are also dealt with; 3. benchmarking of the deployed OS solutions, through a statistical and cost/benefit analysis. Financial, economic, reliability, effort, cost, and time aspects will be considered and integrated; 4. building a European knowledge and experience repository by comparing and pooling knowledge acquired in the previous phases of the project. The knowledge base will be placed on the Internet and made freely accessible; 5. dissemination of the results and the experiences of the project through the knowledge base and a series of workshops at regional and European level.

3.2 Pilot Projects Activity 2 is the crucial phase of the project. Its objective is to run experiments on the introduction of OSS in the partner PAs, and to benchmark the effectiveness of the deployed OS solutions through a statistical and cost/benefit analysis. The analysis will consider financial, economic, reliability, effort, cost, and time aspects.

Data on usage and satisfaction will be collected in the partner PAs by the universities, both manually and automatically. The automatic data collection of process and product metrics (Humprey, 1997 and Fenton and Pfleeger, 1994) is carried out using the PROM tool (Sillitti et al, 2003). PROM allows a fully automatic and non-invasive extraction of product and process metrics, such as time spent on editing a specific document, or effort spent on a project. In that way it is possible to set up comparative experiments on the usage of proprietary and OS software, in order to quantitatively characterize the transition to OSS. The results of the pilot project phase would also be of great interest to OSS developers, as they could check how their applications and tools perform in a corporate environment.

3.3 Observers In order to increase dissemination of knowledge and to promote best practices in Public Administrations, the Consortium has established the role of observer. An observer can access the project’s results and experiences in a privileged way. It may also attend project meetings and thus give useful advises on the implementation of the project itself. At the time of writing COSPA has more than forty observers, among which the University of Alberta (Canada), Victoria University of Wellington (New Zealand) and UNESCO. New observers may join COSPA at any time. On the project’s website (COSPA) it is available the application form for becoming a COSPA observer.

CONCLUSION Successful OSS seems to be so far confined in the system software area, of which Linux, Apache and Sendmail are notable examples. Desktop OSS does not seem share the same amount of popularity. We reported about a trial transition to OSS for desktop applications in local authorities. The success of the experiment has proved that use of OSS on the client side is indeed possible, as recognised by other experiences. However, we had no measures of the effort, cost and benefits of such a transition. A quantitative characterization of the process would be helpful for Public Administrations’ managers, in order to establish the advantages, drawbacks and feasibility of a transition. The issue is addressed at European level in an ongoing research project..

ACKNOWLEDGEMENT The authors thank Dr. Hugo Leiter, Head of the EDP Unit of the Consortium of the Municipalities of the Province of Bolzano-Bozen (Italy).

REFERENCES Netcraft, 2004. http://www.netcraft.com/survey/ FLOSS, 2002. Free/Libre/Open Source Software: Survey and Study. June 2002. http://www.infonomics.nl/FLOSS/ MIGOSS, 2003. The IDA Open Source Migration Guidelines. Interchange of Data Between Administrations (IDA), November 2003. http://europa.eu.int/ISPO/ida/export/files/en/1781.pdf OpenOffice.org project. http://www.openoffice.org Humprey W., 1997. Introduction to the Personal Software Process. Addison-Wesley. Fenton N.E. and Pfleeger S.H., 1994. Software Metrics: a Rigorous and Practical Approach, Thomson Computer Press. Sillitti, A. et al, 2003. Collecting, Integrating and Analyzing Software Metrics and Personal Software Process Data. Proceedings of EUROMICRO ’03, pp 336-342. COSPA project. http://www.cospa-project.org Fuggetta A., 2003. Open source software - an evaluation. Journal of Systems and Software, 66 (2003), 77-90. Raymond E.S., 2001. The Cathredal & the Bazaar, O’Reilly. Pressman R.S., 2004. Software Engineering: A Practitioner’s Approach, 6th ed, McGraw-Hill.