The use of hints in object-relational query optimization
David Taniar, Hui Yee Khaw, Haorianto Cokrowijoyo Tijoe, J. Wenny Rahayu · 2004
There is a growing need to combine traditional relational data with new object feature into one integrated systems called as object-relational database management system (ORBMS) (Dorsey and Hudicka, 1999; Taniar, Rahayu and Srivastava, 2003). ORBMS is built based on SQL3 model, which consist object data types, row types, collection and abstract data types along with basic relational models (Carey, 1992; Stonebraker and Moore, 1996; Fuh et al., 1999; Taniar, Rahayu and Srivastava, 2003). The advantage of database applying SQL3 model is that the system is more flexible and massive scalability so that the immediate changes in the business process can be accommodated fast and easily (Fuh et al., 1999; Taniar, Rahayu and Srivastava, 2003). REF is one of essential data structures which have been introduced in Object-Relational Databases serves as a logical pointer to define the link between two tables (Taniar, Rahayu and Srivastava, 2003). This logical pointer has similar function to pointers in object oriented programming approach. There are three ways of writing REF join queries which are: REF Join, Path Expression Join, and DEREF/VALUE Join. REF Join is the most straightforward method to write Object-Relational queries where the REF data structure is used in the SQL statement to retrieve the REF value (not real data) stored in the ref attribute. Different with REF join, the final result of a Path expression join is the real data taken directly from the referred attribute. Finally, the DEREF/VALUE join method retrieves data from the actual object itself rather than from the table as the other two methods do (Loney and Koch, 2000; Oracle, 2002). Query optimization is the most important stage in query processing where the database optimizer has to choose a query-evaluation plan with minimized cost and maximized