Adaptive Distributed Query Processing.
Yongluan Zhou · 2003
For a large-scale distributed query engine, which supports long running queries over federated data sources, it is hard to obtain statistics of data sources, servers and other resources. In addition, the characteristics of data sources and servers are changing at runtime. A traditional distributed query optimizer or centralized adaptive techniques is inadequate in this situation. In this paper, we introduce a new highly scalable distributed query processing mechanism called SwAP (Scalable & Adaptable query Processor). SwAP can quickly learn and adapt to the °uctuations of the selectivities of operations, the workload of servers, as well as the connection speed without any statistics, and accordingly change the operation order of a distributed query.