Query Processing in Self-Profiling Composable Peer-to-Peer Mediator Databases
Integration of multiple heterogeneous sources is crucial for efficient sharing and reuse of distributed data. An architecture for scalable data integration of many autonomous data sources is presented. In the architecture peer-to-peer (P2P) mediators can be defined in terms of each other through object-oriented (OO) views. Query processing with scalable performance is important to make such an architecture useful in practice. The focus of the described doctoral thesis is on query processing techniques in a composable P2P mediator architecture. Through distributed selective view expansion mediator peers are treated as ‘grey-boxes’ with varying level of transparency. This allows to balance between compilation time and query execution plan (QEP) quality for good overall performance. Self-profiling integrated with the query processor allows for the implementation of adaptive query processing techniques. Adaptive rebalancing of distributed QEPs based on the self-profiling capability of the optimizer detects and re-optimizes sub-optimal QEPs. The proposed P2P mediator architecture and some of the query processing techniques are implemented in the AMOS II mediator system.
Unable to display preview. Download preview PDF.