Well one way or the other the latency of the link does seem to be the most important factor in the problem I'm seeing.
I have three ways to get at the data: 1) log into the remote server using RDP and run the queries locally. even the most complex ones complete in under a second 2) use JDBC over a Bell link with about 75 to 80 ms of delay. most queries take about 1 second, and the bad ones up to 20 3) use JDBC over a Rogers link with 110 to 125 ms of delay. most queries take about 4 seconds, and bad ones take minutes. This is in spite of this link having 6 times the raw throughput. What's interesting is that there does seem to be some relationship between the number of joins and the performance. This makes me *suspect* that the driver is pulling tables to the local client and doing local JOINs.
