Re: histograms, buckets, and bind variables

  • From: Wolfgang Breitling <breitliw@xxxxxxxxxxxxx>
  • To: ChrisStephens@xxxxxxx
  • Date: Wed, 17 Nov 2004 10:49:42 -0700

Bind variables never completely prevented parsing. The first time a sql 
is encountered it needs to be hard parsed. The difference with 9i bind 
variable peeking is simply that at that very first parse, the optimizer 
will take the value of the bind variable into account when deciding on 
an access path. From then on that plan is used. No further peeking 
occurs. That is also part of the problem. If that first parse is done 
for a 1-off bind value, all the other sql are stuck with that plan.

Stephens, Chris wrote:

> This is actually something I've never quite understood.
> 
> 
> ...so with 9i bind variables don't completely prevent parsing correct?
> ...they just make a softer hard/soft parse?=20

-- 
Regards

Wolfgang Breitling
Centrex Consulting Corporation
www.centrexcc.com
--
//www.freelists.org/webpage/oracle-l

Other related posts:

  1. Home
  2. oracle-l
  3. Archive
  4. 11-2004