Hi Lukas, I've used #3 (over #2) and it was a substancial improvement. In any case, I will have a look at #4 in our current environment and tell you what are the results.
cheers, stan. p.s.: i'll be at your JUGS talk :) On Wednesday, May 9, 2012 12:05:08 PM UTC+2, Lukas Eder wrote: > > A follow-up on the previous analysis: > > When running more elaborate tests, ROW_NUMBER() as opposed to ROWNUM seems > to cause quite a heavy performance issue in more sophisticated queries, > which I cannot disclose here. Maybe we should wait with any patching and > benchmark various options in a real, close-to-production setup? > > Cheers > Lukas >
