For the efficient running of a Database Management System there are multiple tuning factors that directly determine its performance. These factors are mostly varied and conflicting and hence adjusting these factors simultaneously for a variety of workload types and highly unpredictable traffic patterns is often a challenge. It has to be tuned based on the variables by a highly skilled administrator. To keep costs of ownership low, increase overall performance of the DBMS and due to the process of tuning being complex, the introduction of fuzzy logic into the structure is necessary by the use of appropriate fuzzy rules to assist in the self-tuning of the DBMS, wherein the control action is expressed in linguistic terms. Besides Fuzzy logic, Neural Networks and Genetic Algorithms can also be used. This significantly improves the query response time. Fuzzy control systems are most suitable to use as they are easy to design, are robust and can be easily tweaked to dramatically improve system performance. It was observed that buffer caches size had a significant impact on the response time with OnLine Transaction Processing type workloads and if appropriately adjusted can boost the response time. Similar results were observed for Shared pool but with a higher value of response time. The fuzzifier module uses fuzzy rules comprising of IF..THEN kind of statements on the fuzzy input variables that decide on fuzzy output variables. This helps in determining tuning parameters. Buffer Hit Ratio and User are examples of such inputs which have significant impact on performance.
S.F.Rodd, Umakant P. Kulkarni and A.R. Yardi (2012). International Journal of Computer Applications. Fuzzy Controlled Architecture for Performance Tuning of Database Management System Article.