A case study in optimizing continuous queries using the magic update technique

The evaluation of continuous queries over data streams often becomes difficult as soon as static context data must be combined with dynamic stream data. This is especially the case if the context data is organized in form of view hierarchies and thus computed from some base facts. In this scenario, typical algebraic optimization strategies fail in providing a well-optimized query evaluation plan which effectively combines the stream and classical view subparts of the given query. The Magic Update method represents a possible solution to this problem as it allows for dynamically generating new selection conditions from the data stream which are pushed into the view hierarchy of context data. In this paper we present a case study in which the performance gain of this technique is shown when optimizing anomaly detection views in an air-traffic surveillance scenario.