Fine-Grained Progressive Algorithm Based on HMJ

Wide-area distribution raises significant performance problems for traditional query processing techniques as data access becomes less predictable due to link congestion, load imbalances, and temporary outages. Non-blocking joining query execution is a promising approach to coping with unpredictability in such environments because of reactively scheduling background processing. Classical non-blocking two-way joining technique based on hash-merge (HMJ), however, fail to deliver acceptable performance in such a scenario where relatively short intermittent delay exists in the gross. We have developed a fairly fine-grained hash-merge join, called HMJ-FG, which has employed a replacement selection tree, allowing many disparted segments to be active in parallel. Using the optimization implementation along with simulation obtained by Tao, we show that HMJ-FG is an effective solution for providing fast query responses to users even in the presence of the longer-term of data sources appeared as unavailability. Theory and experimental results show that our technique delivers results significantly fast under unreliable network.