Adaptive optimization for multiple continuous queries in relational and XML streams

For two different types of data stream — a relational stream and an XML stream, this paper contains an adaptive query optimization strategy called a selection-early. The evaluation order of selection constructs can significantly influence the overall performance of multiple query evaluation. Consequently, based on the filtering capability of the current evaluation sequence of selection constructs dynamically captured at run-time, a selection-early strategy establishes the efficient evaluation sequence adaptively. For this purpose, the overall filtering capability of the current evaluation sequence is periodically monitored and the sequence is rearranged when its filtering capability is varied higher than a specific threshold. Accordingly, this strategy keeps the current evaluation sequence to be as efficient as possible by coping with a dynamic variation of its filtering capability. The experimental studies of the proposed strategy show that it is practically more scalable and stable than other approaches.