A library or consortium with many widely dispersed agencies has the option of installing a single automated system to serve all of its locations, or installing two or more regional systems and linking them with dial-up or leased telecommunications lines. In deciding between centralized and distributed systems, a library or consortium is making more than a technological decision of interest only to systems designers and programmers. The choice between two alternative system configurations will have two major impacts on administrators and users: cost, and query methodology. Normally a centralized system will cost more to operate due to high telecommunications costs, but will provide single-query access to system wide data. On the other hand, a distributed system, while reducing telecommunications costs, may require two or more steps to gain access to all the resources. The assumption underlying distributed systems is that most of the needs can be met within a region.
Under the centralized approach, one or more CPUs are, located at a single site and library holdings and patron information are maintained in a single large data base. All agencies are linked to the central site via telecommunications lines which may be concentrated for economy.
Among the advantages of a centralized system are:
- a single query will provide access to all bibliographic or patron information, regardless of where the material is housed or where the patron was registered;
- only a single copy of each bibliographic record is maintained online; no costs are incurred to store multiple redundant data at two or more locations;
- only one computer room needs to be built, staffed, and maintained;
- only one software license fee, or fewer fees (if the vendor bills on a per-processor basis) is required; and
- the approach reinforces the view of library or consortium resources as a single collection, rather than as a series of separate collections.
There are, however, disadvantages to the centralized approach:
- increased telecommunications costs are incurred because lines are billed by the mile, and must link each location to the central site;
- a very large data base supported with multiple processors is a complex technical environment, more difficult to manage and maintain than several dispersed processors;
- there is diminished overall system reliability because a failure- such as loss of power-at the central site can disable the entire computer system;
- increased CPU costs result from the systems sophistication required to meet the overhead incurred in linking CPUs at the central site;
- increased software costs result from the need to provide multiprocessor linkages; and
- increased maintenance charges are incurred due to the increase in telecommunications equipment.
Under the distributed approach, CPUs are located in two or more geographically dispersed locations. Each processor would support disk drives which store records of bibliographic items located in the region and records of patrons registered in that region. Telecommunications lines would span from each agency to the nearest regional center, and single lines would link the centers for inquiries from one region to another.
Among the advantages of a distributed system are:
- telecommunications costs are reduced because circuits are shorter;
- increased system reliability is achieved because failure at a single site would disable only a portion of the total system;
- increased system reliability results from the use of smaller CPUs in a less complex processing environment; and
- reduced ongoing telecommunications equipment maintenance charges are incurred.
The disadvantages associated with the distributed approach include:
- two or more steps (depending upon the hierarchy selected) may be involved in an inquiry for non-local data;
- increased ongoing maintenance fees are incurred because of the increased disk requirements;
- one-time site preparation costs (electrical, air conditioning, basic construction) and software license fees are increased due to multiple system locations;
- more staff are required to perform
- preventive maintenance and to operate multiple sites; and
- the decision creates a somewhat fragmented system.
These strengths and weaknesses have to be carefully considered before final decisions are made. When the field of potential systems has been narrowed, it is possible to portray the relative cost figures with more precision, based on the prices of the systems then under consideration. While normally cost calculations are made on total five year cost, it is also common to put greater emphasis on comparative operating costs because they compete directly with ongoing personnel and, acquisitions needs.
The more critical issue is that of service to users. Are they well served by providing easy access to the resources of a region and requiring a second step to ascertain resources outside the region? Presumably, were the bibliographic resources of the regions quite similar the answer would be different than were they to be quite different and complementary. A careful analysis would, therefore, include not only technical and economic analyses, but also an examination of the collections and user needs.
