Threshold Values for Processing Distributed Queries
Peter Bodorik, J. S. Riordon · The Computer Journal · 1991
A strategy to process a distributed query is formed using estimates of intermediate result sizes and delays due to network data transfer and CPU processing. Strategy execution is monitored, and corrected if observed parameter values differ substantially from their estimates. This paper presents and compares two methods which can be used to decide when to correct a strategy. In the reformulation method, a new strategy is formulated following each relational operation. The threshold method, on the other hand, is based on the fact that some intermediate results are more ‘critical’ than others. The query-processing strategy is represented as a network of activities, and the critical path method is used to determine threshold values for intermediate results. If an intermediate result is delayed beyond its threshold value, the strategy is corrected. The reformulation and threshold methods are evaluated on a test-bed of queries for a modelled application of a distributed database.