| This dissertation presents QoS management solutions for real-time data services and data stream query processing. The proposed solutions are presented in a stepwise fashion. First, we study data freshness management problem in distributed real-time database systems. For temporal data, we propose to use dynamic replication to reduce the data propagation cost and increase the QoS of data services. For temporal derived data management, we develop a novel QoS management architecture that maintains derived data based on their access patterns. Second, to improve the real-time transaction QoS, we propose a feedback-control-based approach to balance the workloads in distributed real-time databases. In the third part of this dissertation, we investigate how to maintain QoS for real-time queries on dynamic data streams. We develop a real-time data stream query processing system prototype called RTStream. In the prototype system, we extend the query language to support real-time periodic queries specifications and design a QoS management module to dynamically control the data stream admission process and thus maintain the desired system QoS. For queries on dynamic data streams, we employ a novel prediction-based QoS management algorithm that controls the query QoS based on the current stream volume and contents. In distributed environments, we use a similar technique to select the proper propagation schemes for data streams and intermediate query results. In the experimental study, we study the system performance using both synthetic workloads and real workloads from real-time applications such as network traffic monitoring systems. Using our detailed performance studies, we demonstrate that the proposed solutions tolerate dramatic workload fluctuations and still provide desirable QoS for real-time applications.;This dissertation identifies and provides novel solutions to several important research problems in real-time data services QoS support. The key contributions of this dissertation include the dynamic temporal data maintenance algorithms, a transaction QoS management architecture, a real-time data stream query language, a prototype real-time data stream query system and data stream query QoS management algorithms. The research results presented in this dissertation contribute to the state-of-the-art of QoS management in real-time databases and serve as basis for future research in the related field. |
