The main goal of database security mechanisms is to protect the data stored in the database from unauthorized access or malicious actions in general. Typical database security attacks can be classified as: malicious actions executed by authorized or unauthorized users, and an inference attack occurs in multilevel secure database. The focus of this paper is to investigate the effect of malicious transactions detection and association rule mining implementation on database performance. This paper presents implementation of three mechanisms for detection of malicious transactions in the Oracle 10g DBMS and investigates the performance of the three mechanisms using a telephone database, and implementation of association rule mining using (Apriori) algorithm and investigates the performance using Congress Voting Data set. The experimental results showed that the average performance overhead caused by the activation of malicious transactions detection mechanisms is about 40%, and the hiding association rules performance parameters may be very high to ensure sensitive rules hiding.