This month I had the pleasure of speaking at a library conference on the topic of shared catalogs. My role was to describe the virtual catalog approach that underlies the “Kudzu” catalog of a group of libraries in the southeast (see www.library.vanderbilt.edu/iris/vandy.html). Kudzu “covers the south” by connecting the library catalogs of sixteen academic libraries using the OCLC SiteSearch software. My co-presenter, David Singleton, presented a project based on an alternative model where a number of libraries share a common integrated library system, providing a true union catalog that represents their combined collections. Singleton is the Project Manager of the Georgia PINES (Public Information Network for Electronic Services) project (see: pines.lib.ga.us). PINES encompasses over 200 public libraries in Georgia, and is based on a large implementation of the SIRSI Unicorn library management system. These two projects both involve libraries that have joined together in partnership for the purpose of sharing their collections among themselves more efficiently, but they represent fundamentally different ways to accomplish their goals. In this month's column, I will compare and contrast the virtual catalog and the shared library automation system as alternatives for sharing resources among libraries and offering library users access to an expanded set of resources.
In this time of narrowing resources and rising expectations, libraries face many challenges, with many pressures to operate efficiently. The demand to provide more electronic resources, for example, often results in a need to find creative ways to get the most out of our physical collections.
More and more often, libraries band together in partnership so that they can share each other's collections. Today, libraries tend to borrow from each other more so that they can buy less. Whether it be through a formal consortia or an informal arrangement, a group of libraries can work together to find ways to operate collectively and gain new efficiencies. An online catalog that can simultaneously search the collections of a group of libraries is an essential component of a resource-sharing cooperative. Through the implementation of such a combined catalog and other resource-sharing arrangements, a library can expand the amount of material available to its users.
Two of the approaches that might be followed for implementing an online catalog that provides access to multiple library collections include: 1) creating virtual catalog that works on top of the existing library automation systems, or 2) having all the libraries migrate to a single library automation system. These two approaches represent the two ends of a spectrum of options relative to effort in implementation and in the level of institutional commitment among the participating libraries. Basically, the question is whether the libraries want to simply link their existing library automation systems together or if they are able to completely combine their library automation efforts in a joint implementation. There are, of course, other possible options. But for the sake of simplicity and space constraints, we'll limit our discussion to these two.
The Virtual Catalog
Libraries with physically separate catalogs can create a combined catalog “virtually” through a layer of additional software. Virtual catalogs make use of the Z39.50 search and retrieval protocol—by now a standard feature available in all library automation systems. In order to create the virtual catalog, a piece of software is needed that will function as a gateway to the Z39.50 server component of each of the participating library's integrated library systems. The gateway would provide the end user with a Web-based interface for entering searches and viewing results and would operate behind-the-scenes as a Z39.50 client, simultaneously communicating with each of the Z39.50 servers in the system. In effect, the software is a Web-to-Z39.50 gateway. The WebZ component of the OCLC SiteSearch is an example of software performs this function, and is what underlies the “Kudzu” virtual catalog mentioned above.
On the positive side, the virtual catalog approach offers several advantages: It is relatively simple, low-cost way to establish a multi-library catalog, and does not disrupt the existing environment. A virtual catalog can be implemented for a group of libraries that each use different brands of library automation systems. In most cases, the virtual catalog does not replace the library's own Web-based online catalog as the primary interface for searching for materials, but offers an alternative that library patrons can use to find materials not held by their own library.
A virtual catalog can offer a nice variety of features, including display of titles held by each library, including information about multiple copies and holdings. Display of the circulation status of items retrieved in a search is important. Seeing that the book that I need is checked out in one library but available in another is useful information. Just knowing that another library has the material needed is only half of the process. A multi-library catalog also needs to provide a way for the user to get access to the materials in the remote libraries. A virtual catalog interface should also include linkages into an interlibrary loan system, allowing users easy options for requesting materials.
There are, however, some difficulties and complications associated with the virtual catalog approach. Results from queries may not be entirely predictable. Z39.50 is a complex protocol for search and retrieval, offering many features and options in the way that can be implemented. One of the major challenges involved with Z39.50 lies in configuring clients and servers in a diverse environment so that they all agree as to what search attributes, fields and indexes should be used for each search type. Problems can arise, for example, when libraries make different choices about which MARC tags are to be used when performing an author, title, or subject search. Fortunately significant progress has been made in this area by developing profiles that establish consistent ways of implementing Z39.50 in a way that will yield more consistent results.
As the number of libraries that participate in a virtual catalog increases, maintaining good performance can be difficult. The communications model of the virtual catalog involves maintaining a connection between the gateway and each of the target libraries during each search session. The overall search process can only be as fast as the slowest-responding system. Gathering search results from a large number of remote systems is inherently more complex and prone to performance bottlenecks.
Creating virtual catalogs, for the most part, is an approach that works and has been successfully implemented by a number of library consortia. The problems mentioned are generally not insurmountable, but are challenges that need to be addressed.
Share System Union Catalog
A more ambitious approach in providing a multi-library catalog is for the all the libraries to make use of a single library automation system. It is becoming increasingly common for libraries in a district, state, or because of some other common association, to purchase a library automation system as a consortium. Since each of the records reside in the same database, providing the capability to search across all the library's collections is easily accomplished. This model of searching is more straightforward, avoiding the complex matrix of connections involved in the virtual catalog model. The shared system approach for establishing a union catalog should generally offer higher reliability and faster performance.
Implementing an automation system by a group of libraries offers a number of efficiencies beyond that possible when each library procures a system independently:
- Software Price. When licensed as a consortium, the cost per library can be significantly lower than what each library would pay in an independent arrangement.
- Hardware Cost. Especially for the server, a consortium can acquire a large-scale system with many levels of redundancy, and achieve a much higher level of performance and reliability. The storage, memory, and processing power can be scaled in incremental amounts according to the anticipated demands of each additional participating library.
- Maintenance Fees. System vendors usually base the maintenance fees that a library customer pays on a percentage of the initial purchase cost. Given the lower cost per library for the initial license, ongoing maintenance fees will likewise be smaller.
- Technical Staff. Sharing a central system often relieves each individual library from having to hire expensive technical staff such as Unix system administrators, database managers, or programmers. A single team of technical staff can support the central system, greatly mitigating the technical staff that needs to be provided in each library.
- Cataloging. A shared system affords the ability to avoid duplications in bibliographic records, which translates to savings in cataloging efforts. The typical library automation system has a record structure where a bibliographic record describes each unique title, and that copy or holdings statements can be added to each bibliographic record to account for multiple copies owned by the library. One can expect a high degree of duplication—especially in the core collection—of many items among libraries in a consortium. If they each implement a system separately, they would need to create a bibliographic record for each title in the local collection. In a consortium, bibliographic records are shared among the participating libraries. In this arrangement, it is more likely that a new item can be processed by adding a copy statement to an existing bibliographic record. This process of copy cataloging within the consortia is more efficient than having to create or import the higher numbers of bibliographic records that would be required if each library were to operate completely independently.
The fundamental disadvantage to the shared system union catalog relates to the massive investment of staff time and resources in implementing such a system. Especially if the libraries involved have historically operated different library automation systems, the process of procuring and implementing a single shared system can be a very large undertaking. Running a library automation system for a single library is complex enough. Operating a system that serves a large number of libraries presents a significantly larger challenge.
Matching technology to organizational needs
The virtual catalog and the shared system union catalog offer alternative options that can be followed depending on the level of resource sharing that the libraries need to accomplish. The virtual catalog can serve as a tool for increasing the level of resource sharing among a group of libraries. The levels of investment and organizational commitments can be relatively modest. This approach may be best suited for a loose-knit association of libraries.
The shared system model requires a broad strategic commitment among the participating institutions for cooperative resource sharing. Sharing a library automation system forces the libraries involved to cooperate in almost every aspect of their operation. Large pay-offs, however, can be expected in the overall efficiency in library operations for library staff and in the expanded resources available to library users.
We see trends toward consolidation every industry. I've noted many times in this column, for example, the mergers that reflect this trend among library automation vendors. Larger organizations can advantage of economies of scale not available to small independent ones. Libraries, likewise, may often find that they can operate more efficiently as consortia than as separate libraries. Whether it be through a physically combined system or a virtual catalog of linked systems, a multi-library catalog is an important element that supports these organizational trends.