Short Papers
Machine Learning, QoS, Cloud Computing
Cloud computing provides on-demand services with pay-as-you-go model. In this sense, customers expect quality from providers where QoS control over DBMSs services demands decision making, usually based on performance aspects. It is essential that the system be able to assimilate the characteristics of the application in order to determine the execution time for its queries' execution. Machine learning techniques are a promising option in this context, since they offer a wide range of algorithms that build predictive models for this purpose based on the past observations of the demand and on the DBMS system behavior. In this work, we propose and evaluate the use of regression methods for modeling the SQL query execution time in the cloud. This process uses different techniques, such as bag-of-words, statistical filter and genetic algorithms. We use K-Fold Cross-Validation for measuring the quality of the models in each step. Experiments were carried out based on data from database systems in the cloud generated by the TPC-C benchmark.
Victor A. E. Farias, José G. R. Maia, Flávio R. C. Sousa, Leonardo Moreira, Gustavo A. C. Santos, Javam C. Machado