A comparative analysis between two data models of a database

M. F. GOLOBISKY; A. R. VECCHIETTI

Mendoza (Argentina)

Simposio; Argentinean Symposium on Software Engineering; 2006

SADIO

In this work we present a comparative analysis of two different mappings of an UML data model. For this purpose a commercial ORDBMS leader in the market has been used. The first implementation corresponds to the relational model and the schema was designed considering the SQL’92 standard and normalization rules, the second one is based on the object-relational data model and its design was made using the definitions of the SQL:2003 standard. The main goal has been to compare and evaluate the behavior of the object-relational schema and its proposed mappings against the relational approach. The strategy used has been to execute a set of selected queries 5 times, in different moments and calculate the average elapsed time for them; no flush was made to the database buffer pool among runnings. The results analysis is based on response times, queries cost and execution plans. In some object-relational queries several built-in functions and operations have been used. As a consequence of this work we are proposing some extensions to the relational schema diagram to include the object-relational alternatives (references, arrays, multisets, etc.). In this way, the diagram can show in a single view how the object-relational mappings have been implemented.