FILE ALLOCATION PROBLEM: AN EXPANDED PERSPECTIVE.

In designing distributed database systems an important issue is where to store the various copies of each database. This issue is known as the File Allocation Problem (FAP). This research addresses FAP from a new perspective. The main objective is to examine the impact of incorporating a specific concurrency control mechanism (CCM) into the FAP. CCM is an integral part of distributed database systems. Several mechanisms can be used for concurrency control. Depending on the specific CCM used, the communication flows in a network will vary. In order to allocate data optimally, one must identify the exact communication flows in the network. This aspect has been ignored in past research on the FAP. In this research a nonlinear integer programming model has been formulated for the FAP. The model incorporates the central node locking mechanism for concurrency control. The model has been solved using an algorithm called ZOOM/XMP. An analysis for various configurations of a network has been carried out. Assumed values have been used for the various non-decision parameters that need to be fed into the model. Several practical implications have been identified that give insight into the FAP.