Strategies for implementing distributed query algorithms over high-speed, bandwidth-on-demand, wide area networks

Because of the trend toward high bandwidth networks, current distributed query processing algorithms need to be modified or redesigned. Existing distributed query processing techniques must be refocused to reduce latency delay, exploit bandwidth-on-demand, handle multicast issues, and reduce the usual transmission and local processing delays. The authors propose a set of strategies designed to help the implementers of distributed query algorithms on high speed, bandwidth-on-demand, wide area networks. A distributed join algorithm using these design strategies is developed. Using an analytical model they show that the time cost for the proposed new distributed join algorithm is only slightly above that of a theoretical lower bound. Thus, they conclude that the algorithm exhibits near optimal performance and should be considered by implementers of distributed database systems on high-speed bandwidth-on-demand networks.