الفهرس | Only 14 pages are availabe for public view |
Abstract Cloud computing is a promising computing model that provides a combination of parallel and distributed computing paradigms. It has the characteristics of on demand provisioning of a shared pool of configurable computing resources as a service. It provides a cost effective paradigm of computational, storage and database resources to users over the internet. Cloud storage is an important service that is provided by the cloud computing system. It provides the data owners with high accessibility, availability and scalability in respect to increasing the amount of their data in cloud repositories. The increasing number of user’s requesting data from deployed virtual instances can lead to increased loads on the cloud data management system. Multiple queries compete for hardware resources causing resource contention with rapidly changing computational properties and efficient concurrent query executions on especially structured data has become an important challenge. Moreover, it offers numerous benefits regarding better utilization of virtual instances by exploiting parallel executions. This thesis proposes an optimized concurrent queries execution to reduce node contention and handle the degradation in the query processing performance. Our proposed approach combines efficient query optimization and scheduling techniques. Furthermore, it considers the virtual instances load control based on database replication to improve the query processing performance and in addition, it involves a feedback loop comprising of observing, planning and responding to any overloaded node during the query execution. The evaluation of the proposed query processing approach is conducted over a real world cloud using the Amazon EC2 infrastructure provisioning service. The results prove a significant benefit with regards to the overall query processing performance |