A Knowledge Representation and Data Provenance Model to Self-Tuning Database Systems

Most autonomic database systems do not explicit their decision rationale behind tuning activities. Consequently, users may not trust some of the automatic tuning decisions. In this paper we propose a rather transparent strategy, that provides feedback to database administrators, based on information extracted from the database log. The proposed approach consists in transforming log results into a user-friendly knowledge representation, based on the graphical representation for OWL. This model provides users with the rationale behind system decisions, adds semantics to the database self-tuning actions, and provides useful provenance information about the whole process.