As I follow the library automation arena, one of the trends that I notice involves the movement toward more libraries cooperating to share automation systems. Libraries have a long history of cooperating in order to provide improved services to their users and to reduce their costs. In my area of interest - library automation - consortia and other cooperative arrangements provide opportunities to deliver access to broader collections to library users at lower costs to participating libraries.
Not Bound by Technology
More than any other time before, we don't really have to worry about the constraints of possible hardware configurations when thinking about the limits of a library automation installation. We exist in an era of almost infinitely scalable computer platforms. In previous times, we rightfully worried about not building systems beyond the capabilities of the hardware and software. We did lots of calculations to be sure that the number of bibliographic records and their associated indexes would fit within the limits of the disk storage available, that the number of transactions would not overwhelm the speed of the processors, and that the footprint of the software would fit within the memory available.
We felt those limits most during the days of mainframes, when fairly modest computers cost hundreds of thousands of dollars. With each passing generation, the capacity of the hardware has increased dramatically as its costs have continued to go down.
I'm especially struck with this rapidly diminishing cost and increasing capacity of disk storage. In working with the Vanderbilt Television News Archive, we're always in need of more storage for a digital collection that totals more than 120 terabytes (TB). I recently purchased some disk drives offering 1.5TB of storage for about $160, netting a cost of 11 cents per gigabyte. Processing power and memory have increased in similar proportions. While managed storage, including all the equipment needed to house the drives and make them available in reliable and fault-tolerant configurations, increases the cost, the net storage costs can still be as low as 50 cents per gigabyte.
Though the cost of computing power has dramatically decreased, the capacity of any given server remains finite. However, the ability to combine many discrete servers together into clusters results in computing platforms of incredible capacity. Dozens, hundreds, or thousands of discrete computers can form clusters that can handle the computing requirements of the largest organizations. Google, for example, relies on the clustering of low-cost computers for its global search technologies and applications. Although the organization keeps the specifics of its technology secret, in 2006 it was estimated that Google's clusters consisted of more than 450,000 servers (http://highscala bility.com/google-architecture). I mention this figure only to reinforce the notion that it's possible to build computing platforms sufficient for the scale of anything that might be needed in the library realm.
Regardless of the power and capacity of the hardware on which it resides, software may not be able to handle databases past a certain size and can only process a set number of simultaneous transactions. For large-scale implementations, it's important to make use of enterprise-class software that is specifically designed to handle a high volume of transactions and to be able to take advantage of clustering and other performance-enhancing technologies.
A Spectrum of Collaborative Options
We can rest assured that from a technology perspective, the sky is the limit as far as the sheer size of a library automation system or resource-sharing environment. As I look at the way libraries currently implement automation systems, I see options ranging from single stand-alone systems to consortia.
Stand-alone automation: The simplest scenario involves a single library implementing its own automation system. Such a system contains only the bibliographic records of its own local collection, and it manages a patron file of only its own local users. The advantage of this system lies in its simplicity. It's perfectly clear that if a patron finds an item in the catalog, and it's not already checked out, it's available from this particular library. This approach also gives the library the maximum amount of local control. It can configure and customize the system according to its own needs and preferences. The disadvantage of the stand-alone system lies in the limitation of resources that can be presented to the library's users and in the higher costs that will likely be involved in operating the system.
I observe a trend toward fewer single stand-alone library automation implementations in favor of various levels of shared systems. In previous times, the only affordable means for small, often rural, public libraries to automate involved inexpensive PCbased systems running automation products such as Winnebago Spectrum, Athena, Follett, or Circulation Plus. These products have become obsolete, and many of the libraries that use them are moving to shared systems rather than replacing them with another stand-alone system.
Multibranch systems: Libraries with multiple branches usually share an automation system. The standard approach involves an automation system shared among all the branches of a multibranch library organization. In this case, users search a combined catalog that includes the holdings of all the branches and must pay attention to whether the items displayed are available in their local branch. Online catalogs usually have a way to readily identify the items available in the user's preferred branch, with an option to limit results to that location. Many, if not most, multibranch public libraries provide a service to deliver requested items to the patron's home branch.
I also place federated or regional library systems into basically the same category as multibranch libraries. The difference is that while multibranch systems tend to be independent libraries that voluntarily join together, federated or regional systems usually form out of some common funding or organizational structure. Although formed differently, these systems tend to share an automation system in much the same way as branches of a municipal system do.
Sharing an automation system provides library users with access to the combined holdings distributed through all the branches. Since all the libraries belong to the same organization, they usually operate with consistent policies and perceive themselves as all serving the same community.
Consortia: Consortia can offer economies of scale not available to individual libraries. This allows consortia to regularly purchase content products, books, and a variety of other products and services on behalf of their members and to pass on considerable savings.
Many consortia operate library automation systems for the benefit of their member libraries. These consortial systems combine the holdings of several library systems, each of which may have multiple branches. As more libraries participate in a consortial system, the combined holdings produce an ever larger pool of resources. These larger combined collections provide access to a much larger body of material than any of the libraries might be able to present to its users individually.
Because of the geographical distance spanned by consortia, most offer a delivery service where items can be requested and picked up at the user's preferred library. In fact, many libraries have implemented direct consortial borrowing systems that automate the process: A request initiated by the user through the library catalog generates a request to the library that holds the item to route it to the location specified by the patron.
The number of individual libraries gaining access to a library automation system through a consortium is growing. There has also been some consolidation among library consortia, where two or more that previously operated their own shared automation systems have combined into a single, larger system. In recent years, for example, Illinois' River Bend, Heritage Trail, and Northern Illinois systems combined to form the Prairie Area Library System.
Interestingly, I also note some movement in the other direction. I am aware of some libraries that have withdrawn from a consortium to implement their own independent system. As I have talked with libraries that have exited a consortium, the issues tend to involve either lack of sufficient local control, differences of opinion in priorities and strategies, or (surprisingly) excessive costs, perhaps resulting from other services bundled into consortium membership. Rarely do the issues relate to the technology involved in the shared system.
One of the questions that I have been contemplating recently involves how large of a scale of library automation systems can be deployed. Consortial systems have been commonplace for years. But is it possible to achieve greater efficiency and resource-sharing capabilities across even larger geographic regions?
I am aware of a few statewide (or near statewide) projects that have been around for a while. One of the largest that I've identified is the INFOhio project, which provides library automation for all the public schools in the state of Ohio. Through a distributed system of 23 regional centers, INFOhio provides access to SirsiDynix Unicorn for each of Ohio's more than 2,400 school libraries.
Wyoming also practices automation on a statewide level. The WYLDcat, or Wyoming Libraries Database, provides access to a shared SirsiDynix Unicorn system to all the public libraries in the state, as well as most of the public college and university libraries and many of the K-12 libraries. Other states, including Hawaii and Rhode Island, also have statewide systems in place.
In the library arena, we can look at OCLC WorldCat as the largestscale implementation of a collaborative system. OCLC currently posts statistics indicating that WorldCat represents 69,000 libraries spanning 112 countries and holds more than 110 million bibliographic records (see www.oclc.org/worldcat/statistics). The role of WorldCat as a bibliographic service and end-user discovery platform doesn't quite compare to the computing involved in a library automation system, but it gives us some idea of the possibilities of large-scale systems in support of libraries.
Large-Scale Open Source Projects
These statewide automation projects have been in place for some time and are based on traditional, proprietary library automation systems. An emerging trend, however, seems to favor large-scale systems based on open source library automation systems.
The PINES (Public Information Network for Electronic Services) project in the state of Georgia involves providing the open source Evergreen automation system to 52 library systems comprising 277 individual facilities throughout the state. PINES primarily serves the smaller libraries in the state. The larger population centers such as those in Atlanta/Fulton County and Cobb County continue to run independent systems.
The PINES consortium began in the late 1990s as a new consortial effort based on Unicorn. The need to move from an existing system prompted an all-at-once strategy for the implementation of Evergreen.
The Georgia PINES project seems to have sparked interest in other states in moving toward large-scale, if not statewide, automation projects based on open source automation systems. Those that have followed, however, have opted for a more gradual deployment strategy.
Evergreen Indiana, for example, is in the early phase of a project to provide access to a shared Evergreen system for any public libraries interested in participating. Only a few of Indiana's libraries have gone into production so far, but many others are in the queue for future migration.
Alarge portion of libraries will likely make changes over the next few years to implement a new automation environment. While some may choose other alternatives, it seems almost inevitable that the majority will move to some type of shared system, either through a consortium or as part of one of the emerging statewide programs. This movement toward libraries automating through larger-scale implementations will include both open source and proprietary systems. Either way, this reshaping of the library automation landscape should result in lower costs for libraries and access to larger collections of resources for patrons.