Dynamic performance management for database applications in virtualized environments

With the development of virtualization technology, on-demand computing becomes more reliable and acceptable in industry, such as utility computing and IaaS (Infrastructure-as-a-Service) of cloud computing. In virtualized environments, applications are running on VMs (virtual machines) configured with physical computing resources (CPU, memory, etc.). Inevitably, issues about performance management in such virtualized environments are highlighted, such as optimizing the overall performance, balancing the performance of applications running on different VMs, etc. In this paper, we focus on concurrent queries processing in database applications and propose a model-based approach for dynamic performance management in virtualized environments. Based on a normalized performance model we transform the performance management problem into a multiple objective optimization problem and solve it by genetic algorithm. We have carried out some experiments to evaluate the effectiveness of our approach, using PostgreSQL database and TPC-H queries in a virtualized environment.