GregG wrote:Last update on that post 2007, and last version referenced 8.1.7 - so not really trustworthy as a definitive statement. ( http://jonathanlewis.wordpress.com/2006/11/13/trust/ ) But that bit is still true - you can get blocks on disc in a read-only tablespace which are notionally in need of delayed block cleanout, and can't be cleaned.
Thank You, I think I've found something interesting from Mr Lewis :
Looks like those blocks are never cleaned out hence little overhead every read is included .
jgarry wrote:Just a typo - but it certainly sounds as if it ought to mean something. To my ear it sounds like an uncomfortable medical procedure.
11g also moves the goalposts because of its inclinaction to use direct reads anyway on tablescans.Inclinaction: There's a typo that ought to be a real word (if you didn't do it on purpose). Less passive than inclination because something actually happens.