Skip to Main Content
In traditional query processing approaches, execution engine executes query according to an efficient query execution plan generated by query optimizer. However, PDBMS (Peer-based Database Management System) runs in an unpredictable environment, where optimizer is not able to generate an efficient query execution plan based on available statistics. We present a distributed, adaptive data integration architecture-PADIA, which can execute queries quickly and provide query results in an incremental manner. PADIA collects statistics during query execution, and generates different query execution plans for different data segments based on statistics collected. By distributing query executing tasks to different nodes, PADIA lowers work load of single node and accelerates the execution of queries. Experiments show that PADIA is able to provide query results quickly and stably.
Date of Conference: 20-22 Aug. 2010