Report on the Open Source Deployment and Development Conference, 11th December 2003

Document Title File Name File Size Pages Document No. Author: Revision 137239.doc

Last Modification Adrian Stevenson

Report on the Open Source Deployment and Development Conference, 11th December 2003

Contents
1. INTRODUCTION................................................................................................................................3 2. CONFERENCE TOPICS....................................................................................................................3 2.1 Open Source in academia – where are we now?........................................ ....................3 2.2 Development: How Do You Make an Open Source Project?..........................................5 2.3 Deployment: Making the Institutional Case.................................................................... .5 2.4 Deployment: Practical Approaches and Support.................................... .........................8 2.5 Does Open Source Matter?................................................................................ .............9 3. REFERENCES...................................................................................................................................10

137239.doc

2 of 10

09/02/2008 5:54 a2/p2

Report on the Open Source Deployment and Development Conference, 11th December 2003

1. Introduction
The Open Source Deployment and Development conference held at The University of Oxford on 11th December 2004 was the inaugural conference of the JISC funded OSS Watch Open Source Advisory Service [1]. The aim of OSS Watch is stated as follows: “OSS Watch provides the UK further and higher education community with neutral and authoritative guidance about free and open source software, and about related open standards. Specifically, it offers: • • • • A web-based clearing-house for up to date information; conferences and workshops; focussed assistance for institutions and software projects considering open source; investigative reports.”

The agenda covered the following topics: • • • • • • Open source in academia - where are we? Making the institutional case How do you make an open source project? Practical approaches and support Getting the right licence Does open source matter?

This report briefly summarises the issues discussed on the day.

2. Conference Topics
2.1 Open Source in academia – where are we now?
Open Source Deployment and Development - Sebastian Rahtz, Manager OSS Watch and David Tannenbaum, Researcher, OSS Watch The OSS Watch has been set up to provide institutions with neutral and authoritative guidance about free and open source software including any problems using open source. OSS Watch aims to give practical support via its website, focus groups and conferences giving IT managers, project developers and users advise. The OSS Watch does not aim to persuade people to adopt open source or provide a software repository. The OSS Watch have recently completed a scoping study looking at the current situation regarding the adoption of open source in the UK in general, as little was known previously. The number of respondents was fairly small with 34 from HE and 34 from FE. Some general results of the survey were that: • • • • 88% of FE institutions reported that very few members had OSS skills. 59% of HE institutions reported that staff had moderate to significant OSS skills. Current and planned deployment is most significant for networking systems (such as the widely adopted Apache web server) and operating systems (Linux). 25% HE and 53% FE identified cost as the most important reason for choosing OSS.
3 of 10 09/02/2008 5:54 a2/p2

137239.doc

Report on the Open Source Deployment and Development Conference, 11th December 2003 • Interoperability due to open standards is another major attraction.

The main concerns for OSS were: • • • • • Interoperability and migration (OSS not necessarily being open standard). Third party support. Identifying relevant software. Legal, licensing issues – most institutions don’t have a licensing policy. Archiving and managing software.

Some recommendations were: • • • • • More data needs to be compiled on OSS use and on the Total Cost of Ownership. UK and EU policy on OSS and standards should be tracked Training workshops on OSS deployment should be available The development of case studies of institutional strategies explicitly considering OSS and strategy templates based on these. Institutions should be provided with possible licensing routes with particular focus on OSS licensing for commercial exploitation.

What next? • • • There will be a conference in June looking at OSS support and training. A workshop looking at licensing and IPR will be held in March. A workshop will be held in September aiming to write a ‘road map for deployment’.

Open Source in Academia, uPortal in Context – Jim Farmer, uPortal Project Administrator, Instructional Media & Magic Inc. There are three main reasons for using OSS: • • • Features not available in commercial products Cost containment Community process leads to community benefit – shared research and experience

The uPortal Project believe that the observed unit price of commercial software actually increases with time and units sold through practices such as forced upgrades. OSS may reduce software investment costs. The main costs are annual maintenance and integration with other applications. Some models for OSS support were outlined: • • • • Developer contributions – Columbia’s CuCMS [2] Cooperatives and special purpose organisations – Leeds University’s Bodington VLE Commercial software suppliers – iAssessment, Unicon for uPortal Support firms – Red Hat for Linux

The value of the community aspect of OSS is that it can reduce local development costs. Some limitations are that development and support may not be the first priority and that resolution of issues may be delayed. Some points that arose from the following discussion:
137239.doc 4 of 10 09/02/2008 5:54 a2/p2

Report on the Open Source Deployment and Development Conference, 11th December 2003 • Jim Farmer suggested that uPortal was running at about one fifth of the cost of the commercial solution, Vignette. It was noted that there tends to be less documentation available for OSS and the development time, at least in the case of uPortal was relatively long at about 5 years. Another advantage noted is that the OSS allows developers to better assess the security of the software. OSS can give an institution status e.g. Bodington for Leeds University OSS allows staff to improve development skills

• • •

2.2 Development: How Do You Make an Open Source Project?
Bodington Open Source VLE – John Maber Bodington [3] is a web application server suited to the construction of a VLE that is flexible and accessible. It was developed at Leeds University and originally conceived just for local use. After several years of local success the project was pressurised into commercialising. This was considered to be a failure. In 2001 it was decided that the project would go open source. Without giving the matter a great deal of consideration, the Apache Foundation [4] license was adopted with some modifications. The license protects the University’s good name, asserts ownership of trademarks and protects against liability. Why go open source? • • • • • Ability to share development outside of Leeds University not experienced at commercial exploitation The software can survive management problems – other players can pick the software Large Universities with resources for development can protect their investment in the event of the originators withdrawing support for the project. Smaller Universities can adopt the software knowing that if the originators withdraw support, there are several big users who are likely to pick the project up.

Some advice: • • • • • • Need senior management support for going open source Go through OSS Watch for licensing model Make sure you are protected from liability Be prepared to deal with contributors – may get lots of email relating to the OSS There is a need for help (from JISC?) as to how to manage the process from small scale to larger scale OSS development without things getting out of control Need to consider the cost/time implications of documenting the software and supporting information requests relating to OSS

2.3 Deployment: Making the Institutional Case
Open Source? No, Open Standards – Brian Kelly, UKOLN Brian Kelly thinks the focus should be on open standards more than open source.
137239.doc 5 of 10 09/02/2008 5:54 a2/p2

Report on the Open Source Deployment and Development Conference, 11th December 2003 Why Open Standards? Open standards seek to provide: • • • • • • Application independence Device independence Wide accessibility Long term access Neutral ownership & avoidance of licensing conditions, patents Architectural integrity

Open standards characteristics: • • • • Owned by acknowledged neutral body Specifications published openly (and freely?) Developments to specifications open to all Platform and application-neutral

Some relevant open standards bodies are W3C, ISO, ECMA, IETF A warning with respect to open standards: • • • • Open standards may not catch on There can be competing open standards Open standards may be too immature for service deployment (RDF?) Open standards developers are human too! Can make mistakes

QA Focus has developed a matrix for the selection of appropriate open standards. Issues that need to be considered are: Formats: • • • Openness of file formats: Ownership; proprietary spec. but openly published (e.g. PDF) Maturity: How mature is the standard? Functionality: Is the standard designed to provide the required functionality?

Implementation Issues: • • Authoring tools: Are authoring tools which support the standard readily available? Viewing tools: Are viewers which support the standard readily available?

Organisational Issues: • • Resource implications: Licensing & staff costs Organisational culture: Readiness to be innovative; preferences for particular environments

Use of open source and open standards can be of great importance to institutions. But: • • • There will inevitably be complexities, different stages of maturity, failures, etc. There is therefore a need for institutions to develop selection criteria This approach has been taken by QA Focus to support JISC digital library programmes
6 of 10 09/02/2008 5:54 a2/p2

137239.doc

Report on the Open Source Deployment and Development Conference, 11th December 2003 • • There is an opportunity for OSS Watch to be involved in similar work for OSS An important aspect will be sharing – of failures and problems as well as successes

Some conclusions: • • • • • Open standards are critical for the development of open, cross-platform, vendor neutral services Open source software may have a role to play in supporting open standards – but proprietary software can also be used with open standards There are other factors to be aware of (user requirements, maturity, …) There is a need to develop a mature approach to selection of appropriate open source applications – OSS Watch has an important role to play Things change – IBM are now the service company supporting open source. You need to be flexible in responding to changing environment

137239.doc

7 of 10

09/02/2008 5:54 a2/p2

Report on the Open Source Deployment and Development Conference, 11th December 2003

2.4 Deployment: Practical Approaches and Support
Open Sauce? Open Sores? Open Saws? Tales from the front line - Paul Browning, The University of Bristol Paul gave some frank and personal views that were not generalisations but quite specific observations. For example, he outlined a number of specific problems that occurred when using Zope on multi-processor box and running Python on Sun Solaris. In both cases this proved to be very problematic. Paul outlined some things to avoid if OSS adoption is to be successful: • • • • • • • • • • Thinking your problems are unique Not networking Not building a community Not having enough development & test server capacity Being 'small minded' - Mine’s better than yours e.g. Java vs. The Rest Having too eclectic a mix of solutions Paper evaluations of OSS – yours or anyone else’s. Best to learn by doing. Being a proud high priest/guru – avoid people who don’t share expertise like the plague. Thinking free software implies no cost of ownership Not budgeting for support and consultancy

Paul believes that everyone from developers to documentors can and should contribute to the process of OSS selection and implementation. He also suggests you need people to ‘lurk’ on the mailing lists. In many cases the mailing list is the OSS documentation. Paul suggests that due to the success of recent OSS underpinned projects, funding is now outstripping the ability to produce software products. One conclusion is that he believes we need to entertain new models. An example given is that by Rob Page, the CEO of Zope. He makes a comparison that the law is free but noone who can afford it goes to court without a lawyer. What we need are the equivalent of open source lawyers. In addition we need to address how we fill the expertise gap and how we build relationships with third parties such as with other projects or through knowledge transfer. TheOpenCD, Quality Open Source for Windows - Henrik Nilsen Omma, TheOpenCD.org This essentially outlined a project to showcase OSS through the creation and distribution of a CD containing selected high quality OSS. Strict selection criteria were employed in the choice of software: • • • • • • OSI approved license High Quality (stable and 'bug-free') Clean install and un-install Mainstream and easy to use Well documented Best in class

Included on the CD are: • OpenOffice – Word processing and spreadsheet applications
8 of 10 09/02/2008 5:54 a2/p2

137239.doc

Report on the Open Source Deployment and Development Conference, 11th December 2003 • • • • • • • AbiWord – Word processing GIMP – Image processing Mozilla – Web Browser Multimedia applications Utilities The source code OSS literature

The CD specifically does not include any shareware, freeware, servers or development tools.

2.5 Does Open Source Matter?
Does Open Source Matter? - Jeremy Wray, Business Development Executive, Public Sector, IBM Software Group Jeremy began by outlining the essential differences between open source and open standards: Open Standards: • • • • • Interfaces and formats openly documented, accepted by the industry and freely available for adoption HTTP, HTML, WAP, TCP/IP, SQL Typically built by IT companies in consortia – W3C, OASIS The IT industry supports these standards IBM is a major contributor and supporter of Open Standards

Open Source: • • • • Software whose source code is made available allowing anyone to copy, modify and redistribute without fees Linux, Eclipse, Apache, Mozilla, Samba Built by individuals and communities of IT professionals Supported by no-one, except for reputation of backers e.g. IBM, Sun, Oracle, SuSe when they believe it furthers their agenda or that of the market

IBM themselves are fully committed to open source and open standards. They have been committed to Linux since 1999 and have 250 Linux developers. Support was again identified as the key problem for open source software. IBM have been involved in a number of pilot projects with the UK Government to create a spark and get the process of open standards and open source adoption in progress in what was described as a ‘hand holding’ exercise. There are already a significant number of US academic institutions that have adopted the Linux platform. Jeremy then outlined some good and bad reasons for using open source, such as Linux. Good reasons to consider Linux • • • • Competition - Avoids commercial lock-in Security - Avoids security ‘monoculture’ Scalability – When used in clusters & grids Flexibility - Across multiple devices and hardware, allows collaboration between organisations
9 of 10 09/02/2008 5:54 a2/p2

137239.doc

Report on the Open Source Deployment and Development Conference, 11th December 2003 • Potentially makes some projects affordable

Bad reasons to consider Linux • • • Political/religious fervour Because all proprietary software is ‘bad’ Because it’s free

Some general concluding points were that: OSS does not change your needs • • • • License cost typically 5-10% of Total Cost of Ownership Is it fit for purpose? How do I know it will be there tomorrow? Who is going to design, install, maintain, support?

You need to look at OSS with an Open Mind • • • • • Commit yourselves to Open Standards Use OSS as part of a ‘mixed economy’ Does OSS deliver best value for your task? Be a pragmatist - beware the zealots! If it looks too good to be true, it probably is

3. References
[1] OSS Watch Open Source Advisory Service < http://www.oss-watch.ac.uk/ > [2] Colombia University Content Management System (CuCMS) < http://www.columbia.edu/~av317/cucms/ > [3] Bodington Open Source Project < http://bodington.org/ > [4] Apache Software Foundation < http://www.apache.org/>

137239.doc

10 of 10

09/02/2008 5:54 a2/p2