Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> Re: 10046 trace - weird library misses

Re: 10046 trace - weird library misses

From: Tim Gorman <Tim_at_SageLogix.com>
Date: Thu, 13 Feb 2003 13:22:41 -0800
Message-ID: <F001.0054CC7D.20030213132241@fatcity.com>


My comment was not to imply that JDBC Thin is *incapable* of parse efficiency. It's just that if you call "parse" before every "execute" explicitly, even if the SQL statement is exactly the same with only bind-variables values being different, you're going to parse a lot...

One characteristic of JDBC is that it is very low-level, like OCI, in a lot of ways. Lots of folks think that the lower-level mechanisms have some implicit performance advantage, but you really have to know what you are doing to avoid stepping in something foul. In the case of JDBC and OCI, people find it easy not to use bind-variables and to call the same set of SQL operation calls all the time, so they do. And it results in a poorly performing application. They don't realize that the higher-level tools like PRO*C or SQLJ or PL/SQL have optimizations built-in...

Good programmers using JDBC Thin aren't scary; inexperienced or overconfident programmers using JDBC (or OCI or ???) are scary. It's not the tool, it's the programmer...

> Tim - Your statement about the JDBC thin client really grabbed my
attention
> since we plan to create some important systems based on this.
> I thought parsing was related to bind variables (using
PrepareStatement
> in Java), and whether that SQL statement was previously executed and is
> still cached. What am I missing?
> Also we are getting into J2EE if that affects the answer.
>
> Dennis Williams
> DBA, 40%OCP, 100% DBA
> Lifetouch, Inc.
> dwilliams_at_lifetouch.com
>
>
> -----Original Message-----
> Sent: Thursday, February 13, 2003 10:14 AM
> To: Multiple recipients of list ORACLE-L
>
>
> Not an RDBMS bug -- it is the behavior of both your JDBC thin program as
> well as SQL*Plus. Every execute is preceded by a parse in SQL*Plus --
they
> didn't write that utility with parsing efficiency as a primary goal.
>
> Apparently, neither was the JDBC thin code...
>
> ----- Original Message -----
> To: "Multiple recipients of list ORACLE-L" <ORACLE-L_at_fatcity.com>
> Sent: Thursday, February 13, 2003 7:49 AM
>
>
> > Dear friends,
> >
> > I traced one of our test cases and found something weird.
> >
> > Did anybody else observe this?
> >
> > Env:
> > server - 9.0.1.4, Solaris.
> > client - weblogic 7, uses original oracle thin 9.0.1 jdbc driver to
> connect.
> > In fact, I can reproduce all this from SQLPlus
> >
> > Here is an excerpt from tkprof below - why every parse is a hard parse?
> > Looks like the problem doesn't appear when 10046 is not set, and it
appers
> > ONLY on pl/sql blocks returning data to client, normal selects OK. Looks
> > like bug again. Any workaround?
> >
> > And what are these "Misses in library cache during execute"?
> >
> > 9.2.0.2 on Linux works fine, i.e. no misses once it has been parsed.
> >
> > BEGIN :1 := FN_GET_STATUS_ID(:2,:3); END;
> >
> >
> > call count cpu elapsed disk query current
> > rows
> > ------- ------ -------- ---------- ---------- ---------- ----------
> > ----------
> > Parse 40 0.07 0.08 0 0 0
> > 0
> > Execute 80 0.62 1.55 64 1492 0
> > 80
> > Fetch 0 0.00 0.00 0 0 0
> > 0
> > ------- ------ -------- ---------- ---------- ---------- ----------
> > ----------
> > total 120 0.69 1.63 64 1492 0
> > 80
> >
> > Misses in library cache during parse: 40
> > Misses in library cache during execute: 40
> > Optimizer goal: CHOOSE
> > Parsing user id: 40
> >
> > This select
> >
> > select LOADED_VERSIONS, EXECUTIONS, LOADS,PARSE_CALLS,
parsing_user_id
> > from v$sql
> > where sql_text like 'BEGIN :1 := FN_GET_STATUS_ID(:2,:3); END;';
> >
> > gives out whole bunch of these record groups
> >
> > LOADED_VERSIONS EXECUTIONS LOADS PARSE_CALLS PARSING_USER_ID
> > --------------- ---------- ---------- ----------- ---------------
> > 1 1 1 1 40
> > 1 1 1 0 40
> > .... repeated N times
> >
> > Thank you for you time
> >
> > Vadim G
> > --
> > Please see the official ORACLE-L FAQ: http://www.orafaq.net
> > --
> > Author: Gorbounov,Vadim
> > INET: vadim.gorbounov_at_liberate.com
> >
> > Fat City Network Services -- 858-538-5051 http://www.fatcity.com
> > San Diego, California -- Mailing list and web hosting services
> > ---------------------------------------------------------------------
> > To REMOVE yourself from this mailing list, send an E-Mail message
> > to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> > the message BODY, include a line containing: UNSUB ORACLE-L
> > (or the name of mailing list you want to be removed from). You may
> > also send the HELP command for other information (like subscribing).
> >
>
> --
> Please see the official ORACLE-L FAQ: http://www.orafaq.net
> --
> Author: Tim Gorman
> INET: Tim_at_SageLogix.com
>
> Fat City Network Services -- 858-538-5051 http://www.fatcity.com
> San Diego, California -- Mailing list and web hosting services
> ---------------------------------------------------------------------
> To REMOVE yourself from this mailing list, send an E-Mail message
> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> the message BODY, include a line containing: UNSUB ORACLE-L
> (or the name of mailing list you want to be removed from). You may
> also send the HELP command for other information (like subscribing).
> --
> Please see the official ORACLE-L FAQ: http://www.orafaq.net
> --
> Author: DENNIS WILLIAMS
> INET: DWILLIAMS_at_LIFETOUCH.COM
>
> Fat City Network Services -- 858-538-5051 http://www.fatcity.com
> San Diego, California -- Mailing list and web hosting services
> ---------------------------------------------------------------------
> To REMOVE yourself from this mailing list, send an E-Mail message
> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> the message BODY, include a line containing: UNSUB ORACLE-L
> (or the name of mailing list you want to be removed from). You may
> also send the HELP command for other information (like subscribing).
>

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.net
-- 
Author: Tim Gorman
  INET: Tim_at_SageLogix.com

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).
Received on Thu Feb 13 2003 - 15:22:41 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US