Nesterov M. Information Technology for Database Performance Tuning in Business-Critical Systems

The dissertation is targeted on solving the actual scientific and applied task of the development of models, method and information technology capable to provide thresholds that can be used to identify problems associated with the performance of the database and the implementation of successful projects and tuning them. The state of the art in information technology for database performance tuning is performed. The analysis in methods and standards of requirements for development databases business critical systems is presented. The methods and features of information technologies ensuring efficient operating of database were analyzed. Found that most of the modern databases have dynamically configurable parameters that allowing to customizing in real time. However, despite the fact that most database parameters are set to the recommended values calculated using certain rules are not able to maintain the level of performance over the long term. In dissertation developed new method of forecasting the critical states of the database related to the gradual degradation of their characteristics as a result of cumulative effects (overload queries exceeding execution time, etc.). Using mathematical models identified dependency in parameters that supporting stable database state. This information technology ensues increasing the productivity of the database which supports both short-term and long-term forecasts and provides the ability to track various peak metrics. Prediction and comparison with the previous states of database and advance information about the incidents associated with performance reducing the time for further debugging and maintenance, as well as improve the database tuning process.