Logical Foundations of Continuous Query Languages for Data Streams

Data Stream Management Systems (DSMS) have attracted much interest from the database community, and extensions of relational database languages were proposed for expressing continuous queries on data streams. However, while relational databases were built on the solid bedrock of logic, the same cannot be said for DSMS. Thus, a logic-based reconstruction of DSMS languages and their unique computational model is long overdue. Indeed, the banning of blocking queries and the fact that stream data are ordered by their arrival timestamps represent major new aspects that have yet to be characterized by simple theories. In this paper, we show that these new requirements can be modeled using the familiar deductive database concepts of closed-world assumption and explicit local stratification. Besides its obvious theoretical interest, this approach leads to the design of a powerful version of Datalog for data streams. This language is called Streamlog and takes the query and application languages of DSMS to new levels of expressive power, by removing the unnecessary limitations that severely impair current commercial systems and research prototypes.