Integrating Quality of Service requirements in a distributed query processing environment

In recent years, a lot of research effort has been dedicated to the management of Quality of Service (QoS), mainly in the fields of telecommunication networks and multimedia systems. This work led to QoS management strategies for controlling multimedia stream delivery under certain real-time constraints. Emerging applications such as electronic commerce, health-care applications, digital publishing or data mining also have requirements regarding the quality of the service, the cost of the service, the quality of the data to be delivered, the accuracy and the precision of the retrieved data. These examples show the need to consider the concept of QoS from a broader perspective, requiring the collaboration of all the distributed system components involved. We address the problem of pushing QoS into database management systems.; We propose an approach to integrate user-defined QoS requirements, together with the dynamic properties of the system components involved, into a distributed query processing environment. In this approach, the concept of user class is incorporated and users quality preferences are taken into account during distributed query processing. This approach enables the users to give guidance on the QoS dimensions they care about and the tradeoffs they are willing to make. Different QoS dimensions are mapped to various optimization goals while processing distributed queries. The tradeoffs are represented by the weights assigned to each optimization goal. The overall optimization goal is achieved by first applying utility function to each goal and then using weighted sum to the derived utility values.; We are working in the context of multidatabase systems. We follow three steps for global query processing. A series of query processing algorithms are developed consequently. The first step deals with global query decomposition in case of data duplication. Accordingly, we propose a new algorithm for this step, which enables us to consider the server QoS factors (e.g. server load and server availability) in this step. Moreover, the decomposition is heuristic-based and cost-based. The second step decides the join ordering, which can be further divided into two sub steps: left deep tree generation and bushy tree generation by tree transformation rules. Two algorithms are proposed to extend existing approaches accordingly. Last, locations to perform the joins are decided and annotated for the query plan by tree traversal algorithm. In line with different QoS dimensions, several cost models are offered accordingly. In particular, we focus on those QoS dimensions belonging to the performance category (such as response time) and the dollar cost category (such as service charge). The overall cost (in terms of utility) is calculated based on the weighted combination of different cost models. Some experiments are conducted to confirm the effectiveness of our approach.; In parallel with distributed query processing, the problem of data distribution is also revisited and enhanced with the consideration of user class. This work constitutes a first step for providing data distribution strategies allowing database scalability in e-commerce applications. The data distribution strategies proposed include both replication and partitioning. Through experimentation, we demonstrate that our strategy can provide scalability for the database server when the throughput is the major performance concern.