Analysis of fuzzy with neuro-fuzzy approach to self-tune database system

The Database Management system (DBMS) and other system software are used to store and retrieve large amounts of data in a structured way. The modern applications are generally data intensive and involve frequent storage of data and rapid access to them. To drive such applications and to support highly dynamic user load, the database system needs to be tuned systematically. To tune the database system, the database administrator (DBA) has to continuously monitor the DBMS and several other system parameters so as to enhance the performance of the system, and also ensure proper utilization of system resources. The manual tuning is not only tedious, but it may also result in system instability due to over or under-tuning of system parameters. Hence, the trend in modern database systems is to implement self-tuning mechanisms to relieve the DBA of the tedious task. The control architecture proposed is based on the Fuzzy Logic and Neural Networks. The proposed method has been validated under TPC-C workload and has shown significant improvement in performance. The comparison is also made between Fuzzy tuning and Neural network-Fuzzy tuning. With a self-tuning system in place, results have improved performance, with Neuro-Fuzzy Approach.