الفهرس 
SummaryOnly 14 pages are availabe for public view
 |  | from | 252 |  |  |
| | | from | 252 | | |
Abstract
Object-oriented database systems were introduced to meet the needs of new advanced applications which require high data modelling power and new management requirments which can not be met by the relational data model.the performance of these systems is critical since they are characterized by data intensive applications.This thesis is concerned with enchancing the performance of object-oriented database systems>The work in this thesis belongs to two major areas>The first area is query processing and optimization techniques for object oriented database.An Analytical cost model which calculates the cost of processing the querym using several query processing techniques and takes the clustering factor into account is presented.Furthermore heuristics are proposed for the query optimizer in order to cut the search space of possible query evaluation plans.
The second area is solving the physical database design problem by the efficient structing objects in physical storage.Extensions to cuurent clustering techniques for objects are presented.The clustering configuration definition is introduced and two different approaches are developed in order to generate one for a specific object base.The first approach is based on the information about the queries conducted upon the objects in the system.It partitions the object base into highly correlated partitions where each presents a logical cluster and assigns a clustering technique for each.
On the other hand the second paragraph approaches generates the optimal clusteringconfiguration for a certain path in the object base schema graph based on the analytical model presented.The algorithm splits the path into several paths and allocates for each path the best clustering technique and consequently the suitable query processing strategy.