Performance evaluation of database design approaches for object relational data management

Abstract

The output of the conceptual database design may be a relational database, universal relational database, binary relational database, object database or object relational database. There are also a number of different ways to achieve these different database solutions. The performance of the database in terms of storage and query processing is greatly influenced by the database design strategies or approaches. It is difficult to find a single strategy to achieve good performance in all types of the database applications. This thesis presents a number of approaches for the design of relational and objectrelational databases. We have designed a number of test cases to evaluate the performance of the design approaches and to evaluate the performance of the database. On the basis of different data models, analytical evaluation is performed for all the design approaches. Experimental evaluation for all different approaches has been performed by using both synthetic and real data sets with different characteristics. The main components that we have considered for different approaches are denormalization, nested relations and object orientation. We have developed an algorithm for denormalization, based on SQL query frequency in different modules in an application. We have applied this algorithm in real as well as synthetic data sets and found effective in schema denormalization. We have evaluated universal, binary, object and object related schema for different applications cases and found that relation schema offers moderate performance in terms of both storage and response time.