One of my main professional interests involves following the library automation industry. I maintain a Web site devoted to this topic and regularly write about the companies and systems that comprise this arena. So it is with great interest that I consider what impact Open Source software might make on this industry. The Open Source movement could effect radical changes to libraries should it produce an integrated library system that earned a level of acceptance on the same order that Apache did in the Web server market. Like Apache, an Open Source ILS would have to offer top-of-the-line features and performance to gain acceptance over its commercial rivals.
My general approach to software, technologies, and systems is initial skepticism. I've learned that the initial hype about any new technology usually exceeds its practical impact in the long term. My attitude toward Open Source software in libraries is no different. While I appreciate its successes, I also recognize its limitations. There is no doubt that Linux and Apache represent a worldwide victory over high-powered commercial opponents in the operating system and Web server arenas. (1)
I do not, however, expect to see such victories of Open Source software over commercial products in the integrated library system arena. Both broad historical and recent trends argue against a movement toward libraries creating their own library automation systems—either in an Open Source or closed development process.
An undeniable trend in the course of library automation involves a movement toward vendor-supplied systems and away from locally developed ones. Libraries large and small recognize that they do not have the resources to develop and maintain library automation systems. Some of the recent examples that come to mind include:
- Library of Congress: Adopted Endeavor Voyager to replace several locally developed systems.
- UCLA: Early implementer of DRA Taos over their locally developed ORION system.
- Stanford University: Abandoned locally developed BALLOTS system for SIRSI Unicorn
- Penn State: Converted from their locally developed LIAS system to SIRSI Unicorn
Other less prominent libraries that have left locally developed systems for commercial systems include Carelton University, Jefferson County (CO) Public Library System, the Fogelson Library of the College of Santa Fe.
Very few large libraries continue to operate locally-developed library automation systems. The only two ARL libraries currently running locally developed systems are the University of California at Berkeley and the University of Texas at Austin. Neither of these locally developed systems is Open Source. It should be noted, however, that the development of these systems far predates the Open Source movement. One may well speculate that had the Open Source movement been in place during the period in which many libraries were creating library automation systems, the current environment of reliance on commercial systems would be quite different.
The complexity of Library automations systems exceeds the pool of available volunteer programmers. A full-fledged ILS can easily be a software system of a million lines of software code. Library automation companies that have recently undertaken the development of a new ILS have generally expended about 5 years of development time with a team of 30-50 programmers. The creation of a new ILS is a multi-million dollar project. It is hard to see that even a large collective of libraries would have the available programming staff to develop and maintain a large-scale ILS.
Preferred technology architectures evolve faster than the development cycles of applications built upon them. The history of technology has seen constant shifts in the computing models and architectures. Time-sharing host/terminal systems gave way to client/server systems with thick graphical clients. These gave way to an n-Tier architecture and Web-based clients. Some models prove to be more long-lasting than others. Recent events in the industry bring the ill-fated Taos system to mind as an example. CORBA and Object-Oriented databases, the underpinnings of Taos that were in the height of fashion when DRA's Taos system was conceived, do not currently enjoy the same level of respect. Even with massively funded, concerted development efforts by commercial companies, a system is doing well if its technical infrastructure is not obsolete by the time the application is fully developed.
Even with their teams of full-time development and support personnel, the commercial companies often cannot keep up with their library customer's expectations. Is it realistic to believe that a cadre of Open Source developers would have the available resources to keep pace with the ever-rising expectations of libraries for system enhancements and ongoing support?
The Open Source approach would posit that a world-wide collaborative effort would bring about more satisfactory software since it would be built for and by libraries. Yet it is hard to see that teams of programmers, distributed through libraries all over working in a mostly voluntary mode would have the time allotment, project management infrastructure, and other resources needed for the concerted development efforts required to build and maintain an integrated library system.
Libraries expect far more from their library automation vendors than software. In the current market of relatively mature systems each of which has achieved a high level of core functionality, differentiating factors include quality of support, the vision of the company regarding broad technology and industry trends, guaranteed continued development of the system relative to evolving requirements, standards, and technology architectures.
While there are numerous Open Source enthusiasts in the technology and computer support units of libraries, few library administrators have demonstrated interests in taking on the risks and responsibilities of strategic reliance on Open Source library automation systems. A handful of programmers in libraries currently work on Open Source projects in small blocks of time either officially or unofficially allocated. But it does not seem likely that library administrators are ready to switch from paying license and support fees to commercial firms to paying for local development and support. Programmers are very expensive personnel. Libraries generally lack the ability to fund adequate programming and technical staff in support of commercially-supplied systems, much less toward the development of new Open Source systems.
Most of the current Open Source ILS projects that I am aware of can be characterizes as relatively small-scale efforts. Some rely on paid staff, others depend on volunteers. It is hard to see how a full-fledged scaleable ILS can emerge out of the current projects. The Open Source ILS projects that I have been tracking include: Avanti (www.nslsilus.org/~schlumpf/avanti/). OSDLS / Pytheas (osdls.library.arizona.edu), Koha from Katipo Communications (www.katipo.co.nz), and Open Book from the Technology Resource Foundation (http://www.trfoundation.org/). The latter two systems show the most promise. Yet it is hard to discern any significant trend of even small libraries adopting these systems in favor of the small-scale systems available from commercial vendors. OpenBook, which may show the most promise, is still not ready for primetime, and behind its initial schedule.
Several technical components that are useful to those that build and integrate library automation systems and other library software are available. Some of these Open Source utilities include:
- James, a Java API for MARC records
- YAZ, a Z39.50 toolkit
- ZAP, an Apache module for Z38.50
- XMLMARC, utility for converting MARC records to XML
- ZETA Perl, A Perl module for Z39.50
My concluding observation in the ILS arena is that none of the current library automation vendors have expressed concern that their efforts will be usurped by Open Source efforts. Their customer base continues to grow and they each describe a long-term agenda of future development. Libraries are buying more commercial software, not less. While my idealistic side might like to see libraries able to obtain automation systems without cost in an Open Source arrangement, my practical observations show little movement in this direction. As one who closely follows the library automation industry, I can see no paradigm shift approaching where commercial companies yield to open source and free software.
While I've argued that an Open Source ILS does not seem to be a realistic expectation, at least not in the near future, I do see other significant opportunities for libraries to take advantage of the Open Source model.
It is likely that Open Source will impact libraries in the non-ILS arenas—which may ultimately be more important. The recent interest in digital library initiatives present opportunities for Open Source. The early software for the Open Archives Initiative, for example, is largely Open Source. As XML gains ground in the library arena, it is also likely that a significant body of Open Source software will arise.
We already see some inroads of Open Source software for libraries. OCLC, for example, recently released the Java source code to its SiteSearch toolkit for non-commercial use. While OCLC determined that SiteSearch could not be sustained in the usual commercial development and support model, it is willing to provide some level of help to the community of libraries want to build a future for SiteSearch as a Community Source application (because of the limitation not allowing commercial use of the software, the SiteSearch license is not a true Open Source license). OCLC also released their Pairs search engine, and several other tools under Open Source licenses, and appear to be interested in future Open Source projects.
George Mason University has created OSCR (Open Source Course Reserves), an electronic course reserves system designed for academic libraries (see timesync.gmu.edu/OSCR).
Some commercial companies use some Open Source components within their systems and even contribute Open Source modules. epixtech, for example, indicates that they will develop an NCIP interface in an Open Source license arrangement (2).
The opportunity for libraries to develop Open Source applications in the digital library arena is narrow. One of the major trends among the ILS companies involves an interest in creating products in the realm of broader information delivery, content integration, and resource sharing. It will be interesting to observe whether Open Source applications will be developed and survive in these areas that can compete with the emerging commercial offerings.
For a list of resources related to Open Source software see Library Technology Guides (staffweb.library.vanderbilt.edu/breeding/ltg.html), select the bibliography section and perform a subject search for “Open Source Software.”
References and Notes
- For an article affirming these trends, see Wheeler, David A. “Why Open Source oftware / Free Software (OSS/FS)? Look at the Numbers!” www.dwheeler.com/oss_fs_why.html. Accessed Feb 17. 2002.
- Epixtech, Epixtech Announces Open Source Licensing Plans for NISO CIP "Engine," Jan. 15, 2001. Accessed Feb 17, 2002. www.epixtech.com/ls/press/2001/9998.asp.