Numbers in bug 114696, comments 45 and 46 seem to hint that mysql doesn't do a good job of handling the joins we do in the results. It may be worth splitting up the searchs, so that we select all the required bug_ids, then do inner joins to get the various column headers. Based on those large numbers, this will be requried before we start splitting stuff up into lots of lookups with custfields. However, the disadvange is that for short queries, the extra query may slow stuff down, because of the db->perl->db step. This isn't an issue on databases which support subselects, because they can pull the query out if its worth it. I'm not sure that the CREATE TEMPORARY TABLE overhead would be much better, either. Anyway, I'll try to come up with a patch, and then we can run some tests.

This has been implemented in Bugzilla 4.4.