RE: rm RULE based optimizer != GOOD IDEA

  • From: "Pete Sharman" <peter.sharman@xxxxxxxxxx>
  • To: "Ethan.Post@xxxxxx" <Ethan.Post@xxxxxx>, "Christian Antognini" <Christian.Antognini@xxxxxxxxxxxx>
  • Date: Sat, 23 Apr 2005 01:28:49 +1000

So let me chime in with my personal viewpoint on why this would happen afte=
r I issue the following SQL command:

SQL> SELECT standard_disclaimer FROM company_requirements;

You can imagine the output.

Let me say up front that I haven't read all the messages on this topic for =
two reasons.  Firstly, I'm too damn busy at the moment, and secondly for so=
me reason I'm getting (for some messages, not all) up to 6 or 7 copies of e=
ach message so I'm bulk deleting a lot of them.

There may be some minor improvement in the performance of the RDBMS engine =
by removing the RBO code, and the key word here is minor.  It's only becaus=
e of the reduced code path that you get this minor improvement.  Less check=
ing of which path to go down.

But to me, that's not the reason for swapping from 2 optimizers to 1.  Putt=
ing it bluntly, 2 optimizers mean twice as much work for the developers of =
the kernel.  Any change needs to be checked to ensure it works against both=
.  Given that the CBO DOES largely choose the right execution path now, why=
 would you want to continue to support code that honestly speaking is so ol=
d that probably all the developers that worked on it have retired?

I'm not trying to open this up to an argument on how effective the CBO is. =
 My personal opinion again is that most of the time it works.  So you need =
hints some of the time - who cares?  At least you have the flexibility of u=
sing them, unlike some other DBMS engines that you can't affect at all that=
 way.

 =

Pete
 =

"Controlling developers is like herding cats."
Kevin Loney, Oracle DBA Handbook
 =

"Oh no, it's not.  It's much harder than that!"
Bruce Pihlamae, long-term Oracle DBA
-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] =
On Behalf Of Post, Ethan
Sent: Friday, 22 April 2005 10:57 PM
To: Christian Antognini
Cc: oracle-l@xxxxxxxxxxxxx
Subject: RE: rm RULE based optimizer !=3D GOOD IDEA

I guess my question would be why?  Is the engine going to be faster
because they have removed this bit of code?  Maybe you stop supporting
it but why make it impossible to use as a tool in the future?=3D20

-----Original Message-----
From: Christian Antognini [mailto:Christian.Antognini@xxxxxxxxxxxx]=3D20
Sent: Friday, April 22, 2005 4:22 AM
To: Post, Ethan
Cc: oracle-l@xxxxxxxxxxxxx
Subject: RE: rm RULE based optimizer !=3D3D GOOD IDEA

<snip>

The feature is already gone. In "Oracle Database Performance Tuning
Guide 10g Release 1" you can read:
"The CHOOSE and RULE optimizer hints are no longer supported. The
functionalities of those hints still exist but will be removed in a
future release."


--
//www.freelists.org/webpage/oracle-l

Other related posts: