Quoting "Gogala, Mladen" <MGogala@xxxxxxxxxxxxxxxxxxxx>: > Actually, once upon a time in Lubljana Slovenia, I benchmarked a system > called Wyse 6000 against SGI Indigo2. > Another member of this list was playing for the competition. I did put redo > log files on a RAM-disk and it did make > a lot of difference, but it has happened a long, long time ago (RDBMS > 6.0.36) in a galaxy far, far away. Still made a difference as late as 8ir3, last time I had a chance to test a SSD on the redo logs. Where IMHO it can make a significant difference: if the system does a lot of db writes (update/insert/delete) and it's not practical/economic to dedicate an entire huge disk (EMC raid or not is immaterial) to redo files. Outside of those conditions, pot luck really. Like Gary said: measure and derive would be the way to go. -- Cheers Nuno Souto from sunny Sydney -- //www.freelists.org/webpage/oracle-l