Adaptive Query Processing in Cloud Database Systems

In cloud environments, resources should be acquired and released automatically and quickly at runtime. Thereby, the implementation of traditional query optimization strategies in cloud platforms can have a poor performance, because they cannot predict future availability and/or release of resources. In such scenarios, adaptive query processing can adapt itself to the available resources to run queries and, consequently, present an acceptable performance in response to a query. However, traditional and adaptive query optimizers main objective is to reduce response time. Moreover, in the context of cloud computing, users and providers of services expect to get answers in time to guarantee the SLA. Therefore, we propose a framework that uses adaptive query processing based on heuristic rules and cost of failing the SLA. It will be implemented on structured data, considering that some cloud computing platforms support SQL queries directly or indirectly, which makes this problem relevant.