OPTIMAL REORGANIZATION SCHEDULES OF STRUCTURAL DATABASE DETERIORATION

Both response time and processing time become recently big problems in the processing time of various online and batch systems. Especially, the access efficiency of the database should be greatly controlled as for the speed of the processing time of accesses to the database. Because records are inserted far from their logical position, the storages of deleted records occupy some diverse spaces in the database. Then, to cover the weak point of the database, we execute the database reorganization at suitable times to achieve a good performance requirement for the application. There are two types purposes of the database reorganization: The purpose of physical reorganization is to optimize the database storage and to improve the database structure. However, as the database has usually access locality, its data structure may deterirate in limited parts of the storage space. Thus, we adop the partial reorganization. This reorganizes only locally structurally deteriorated space in the database, while the structural efficiency can be recovered similarly to the full reorganization. This paper considers two structural deteriorations which increase with time and occur independently in time. When the amount of deterioration is estimated at periodic time and at a specified time, the expected cost rates are obtained, using the cumulative damage model, and optimal policies which minimize them are discussed and analytically. We compute optimal policies for two models and compare them numeically.