Re: Designing database tables for performance?
Date: 7 Mar 2007 08:56:03 -0800
On Mar 7, 10:15 am, Bob Badour <bbad..._at_pei.sympatico.ca> wrote:
> Walt wrote:
> > "Bob Badour" <bbad..._at_pei.sympatico.ca> wrote in message
> >>Thus what some folks call logical I/O neither happens at the logical
> >>level of discourse nor does any I/O.
> > Not all participants agree with you about what "the logical level of
> > discourse" is.
> Why not? It has been well-defined since Codd introduced the concept
> during "The Great Debate".
My .02: the semantics of "logical I/O" originally confused the heck out of me - for the longest time, I figured it meant "do nothing" (e.g. results of operation were logically inferred). In my mind, a read from DUAL (the Oracle TABLE_DEE) might be a "logical I/O" - the optimizer would literally read nothing to resolve the contents of DUAL, because the contents exist at a logical level. If it was a physical process, I figured, surely they would call it a "physical <something>" (like "physical i/o" or "physical buffer i/o" or "physical SGA i/o").
In other news, Oracle gurus will be wiping "a hole in the ground" from this day forward. ;-) Received on Wed Mar 07 2007 - 17:56:03 CET