Performance-Aware Refactoring of Cloud-Based Big Data Applications
Chen Li, Giuliano Casale · 2017
The problem of optimizing performance of cloud-based Big Data applications is critical in the cloud computing domain. In this paper, we propose a performance-aware approach that not only considers the cost of the cloud resources but also focuses on the dependency constraints among the deployed components to help dynamically refactoring the application. Furthermore, our approach closes the gap between design time models and runtime models. The feedback can be provided to application designers to iteratively enhance the application design and improve the application deployment. We have experimented our approach on the Wikistats application and the results show that the proposed approach effectively refactors the deployment based on different QoS metrics (e.g., utilization, cost, availability) of the cloud resources and dependency constraints.