| A Streaming Data System is a data processing system that consists of source nodes and sink nodes. Source nodes continuously detect domain-specific events and generate data streams, and sink nodes evaluate user-issued queries over data streams and deliver results to users. Streaming Data Systems are becoming more important because they are able to continuously provide information on currents events in or near real time fashion. This is particularly useful in dangerous environments where people are hard to reach and where tremendous volume of data exists that people are unable to analyze in real-time.; However, due to the infinite volume of data streams and the different semantics of query operations, traditional database systems lack the ability to process streaming data. Recently, a lot of research in streaming data systems has been reported to fill this gap. Nonetheless, most proposed systems aim at specific domains. Even though few researchers have proposed a general architecture for Streaming Data Systems, their general architectures concentrate on only a subset of those systems, such as they query processing system, and have many underlying assumptions that have not been explicitly specified.; In this thesis, our goal is to provide a general model and architecture for Streaming Data Systems as a domain independent model. We specify our model in terms of architecture, data element types, query operations, and functional components. This thesis can serve as a reference architecture to better understand and categorize research in the area of Streaming Data Systems. |
