OPTIMIZED QUERY EXECUTION IN A DISTRIBUTED DATA STREAM PROCESSING ENVIRONMENT

摘要

A controller (140) is described for controlling the processing of a data stream (120) by a data stream processing system (100) comprising a plurality of data stream management systems, DSMSs (110-1 to 110-N), wherein each DSMS is arranged to execute a respective continuous query comprising an operator arranged to operate on windowed portions (120A) of the input data stream to generate an output data stream (130) comprising continuous query execution results. The controller (140) controls the execution of the continuous query on a windowed portion of a data stream when different DSMSs receive different parts of the data for the windowed portion (120A). The controller predicts how the identity of the DSMS receiving the data stream will change based on prior receptions of the data stream (120) by the DSMSs. The controller (140) then selects, for processing of the windowed portion (120A) of the data stream, based on the prediction and the size of a window used to obtain the windowed portion (120A) of the data stream (120), a single DSMS of the plurality of DSMSs that is to execute the respective continuous query. The controller generates a control signal to cause only the selected DSMS to execute the continuous query using data in the data stream received thereby, so that the continuous query is executed on data in the windowed portion (120A) of the data stream (120) only by the selected DSMS.