The Index Suggestion Problem for object Database Applications

The index suggestion problem for an object database management system (ODBMS) application is to automat-ically identify indexes which may speed up the applica-tion. We discuss how index choice for an ODBMS ap-plication differs from relational database (RDBMS) index optimization, present novel parts of an intuitive cost model for index-related ODBMS operations, and describe the re-sults of experiments designed to validate the model. The model is used to find a list of candidate indexes with ex-pected benefits for an example execution of the application with the user’s test data. Our results demonstrate that an intuitive approach can be used to discover and compare access paths which may improve the performance of the application.