Yeah, I agree with Cary and Mark and would add a comment that a tricky thing like changing block size (thinking about granularity of buffer locking) should be tested with simulating real concurrency. E.g. your single session index lookup might run faster due lower index height, but on the other hand you could have more buffer busy waits in high-concurrency environments, etc.. Tanel. _____ From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Cary Millsap Sent: Thursday, January 11, 2007 00:13 To: mark.powell@xxxxxxx; oracle-l@xxxxxxxxxxxxx Subject: RE: Any performance benefits in going to db_16k_cache_size or db_32k_cache_size I have the same opinion as the one Mark describes here. One more comment: Why guess, when you can KNOW. If you need to know, test it, and measure the performance.