| Distributed computer systems are attracting more and more attention as possible complements or replacements to the centralized computer systems. These systems are considerably more complex compared to the centralized ones and require solution of various difficult problems that arise in designing and operating them. This dissertation concentrates on the issues of locating computers and allocating databases among these computers. As of yet, these issues have not been satisfactorily resolved.;The problems outlined in the previous paragraph are formulated as linear integer programming models. Solution procedures are developed for each of these problems. Subsequently, these models are treated as design tools and are used in investigating the implications of various design decisions. The experiments performed for this purpose, illustrate not only the use of these tools but also provide valuable general information about the issues involved in designing distributed computer systems.;Depending on the circumstances, the problem of computer and database location can take on different structures. In many cases, organizational factors and management preferences result in predetermination of computer sites as well as the type of computers to be placed in each site. If this is the case, then the problem becomes one of assigning databases to a set of computers without violating various capacity constraints of each computer. This is the first problem addressed in this dissertation. The second problem pertains to the design of homogeneous computer systems. Here, the designer determines not only the type of computer to be used but also the number and location of computers and the allocation of databases to these computers. The third problem addresses issues faced in designing heterogeneous computer systems. This problem includes decisions regarding the computer sites, size and type of the computer in each site and the allocation of databases among these computers. |
