Re: 9i Parsing Optimization

  • From: "LS Cheng" <exriscer@xxxxxxxxx>
  • To: ax.mount@xxxxxxxxx
  • Date: Thu, 31 Aug 2006 10:42:07 +0200

Hi Alex

I think the improvemente might be cuased by the new OCIStmtPrepare2 call
(introduced in 9.2), this uses a client side sql statement cache to avoid a
server side parsing.

Regards

--
LSC




On 8/31/06, amonte <ax.mount@xxxxxxxxx> wrote:

Hi

I am checking the ODBC version being used.

It is Merant ODBC v4.1 and has not been changed. My doubt is if this ODBC
driver depends internallly on OCI version. If that is the case then it
probably was using OCI8 and now using OCI9 and may be we are talking about
OCI9 optimization.

All I have is a libodbc.sl which is supposed to be the Merant ODBC, how
can I check if this has dependency on OCI?

Thanks all

Alex



On 8/31/06, Anjo Kolk <anjo.kolk@xxxxxxxxxxx> wrote:
>
> Part of the OCI8 interface is the capability to execute SQL statements
> without doing defines (for the select list). So you can do an execute and
> then after that execute you know what the select list looks like (types,
> sizes), for this an implicit DESCRIBE (which requires a parse) is done,
> after that a normal regular parse is done. So that could explain the 2
> parses in 8. May be that was optimized in oracle9. When you did the defines,
> one can Fetch the data.
>
> If I remember correctly (talking to some folks from siebels many years
> ago), they do use OCI (not sure if it was OCI8, but most likely)
>
> Anjo.
>
>
> On 8/30/06, Jonathan Lewis <jonathan@xxxxxxxxxxxxxxxxxx > wrote:
> >
> >
> > I don't know what technology Siebel Tools uses,
> > but your earlier double call may have been a
> > "feature" of the ODBC or JDBC or whatever
> > driver was between Siebel and Oracle, rather
> > than a bit of the application code.
> >
> > I believe there used to be a feature in the Perl
> > driver, for example, that made it do a parse
> > call without bind values, followed by an execute
> > call with bind values - but if the bind types weren't
> > all character, the execute call would require a
> > second parse call.  It's something like that that
> > may have gone away on the upgrade.
> >
> > Don't ask me to quote specifics - it's a long time
> > ago and in another version.
> >
> >
> > Regards
> >
> > Jonathan Lewis
> > http://www.oracle.com/technology/community/oracle_ace/ace1.html#lewis
> >
> > The Co-operative Oracle Users' FAQ
> > http://www.jlcomp.demon.co.uk/faq/ind_faq.html
> >
> > Cost Based Oracle: Fundamentals
> > http://www.jlcomp.demon.co.uk/cbo_book/ind_book.html
> >
> >
> > ----- Original Message -----
> > From: "amonte" < ax.mount@xxxxxxxxx>
> > To: < jonathan@xxxxxxxxxxxxxxxxxx>
> > Cc: <oracle-l@xxxxxxxxxxxxx>
> > Sent: Wednesday, August 30, 2006 9:50 PM
> > Subject: Re: 9i Parsing Optimization
> >
> >
> > > Anjo and Brandon, I forgot to mention we use RULE BASED OPTIMIZER so
> >
> > > permutations shouldnt be a problem. This is not PL/SQL so it cant be
> > some
> > > pl/sql optimization. This is a Siebel Workflow Manager Process.
> > >
> > > Jonathan, it is Siebel Tools, I understand that from application
> > codes you
> > > can invoke two parse calls but what I dont understand is why the
> > second call
> > > is not "called" in 9i?
> > >
> > > TIA
> > >
> > > Alex
> >
> > --
> > //www.freelists.org/webpage/oracle-l
> >
> >
> >
>
>
> --
> Anjo Kolk
> Owner and Founder OraPerf Projects
> tel:    +31-577-712000
> mob: +31-6-55340888
>


Other related posts: