With increasing popularity of fast local area networks, the objective of distributed query processing has shifted to minimising query response time by load balancing and parallel processing. This paper presents an approach to dynamic distributed parallel query processing. The approach attempts to ensure good performance of query processing by delaying some optimisation decisions until queries are to be executed or modifying, during execution, any original decisions that are found to be wrong. The dynamic parameters considered are intermediate results of queries and workloads of processing sites.
展开▼