Why is optimizer different in sql than in pl/sql?

Hello, I am running Oracle database 9i, it is running on an On-Demand hosted server, its purpose is to serve our Oracle Payroll application.
I have a certain query, and first tried it as a stand-alone query. It displayed the results almost immediatly.
Then I inserted the exact same query in a pl/sql block. It took more than one hour and canceled it. By looking at the session browser in TOAD, I saw that it was stuck on that query.
I also saw that the explain-plan for the query changed drastically from running as stand-alone sql to running inside pl/sql.
Why is this happening? How can I avoid this?
Please note that I do not intend to use optimizer hint since the indexes are built by oracle On-Demand and not me, so I don't know which indexes exist.
Here is the explain plan when running from stand-alone SQL:
Operation Object Name Rows Bytes Cost Object Node In/Out PStart PStop
SELECT STATEMENT Optimizer Mode=CHOOSE          1           65
bq. SORT AGGREGATE          1      72
bq. bq. TABLE ACCESS BY INDEX ROWID     HR.PAY_ELEMENT_ENTRY_VALUES_F     1      13      4
bq. bq. bq. NESTED LOOPS          1      72      65
bq. bq. bq. bq. MERGE JOIN CARTESIAN          1      59      61
bq. bq. bq. bq. bq. TABLE ACCESS BY INDEX ROWID     HR.PAY_ELEMENT_ENTRY_VALUES_F     1      13      4
bq. bq. bq. bq. bq. bq. NESTED LOOPS          1      36      33
bq. bq. bq. bq. bq. bq. bq. TABLE ACCESS BY INDEX ROWID     HR.PAY_ELEMENT_ENTRIES_F     1      23      29
bq. bq. bq. bq. bq. bq. bq. bq. INDEX RANGE SCAN     HR.PAY_ELEMENT_ENTRIES_F_N50     44           3
bq. bq. bq. bq. bq. bq. bq. INDEX RANGE SCAN     HR.PAY_ELEMENT_ENTRY_VALUES_F_N50     9           3
bq. bq. bq. bq. bq. BUFFER SORT          1      23      57
bq. bq. bq. bq. bq. bq. TABLE ACCESS BY INDEX ROWID     HR.PAY_ELEMENT_ENTRIES_F     1      23      28
bq. bq. bq. bq. bq. bq. bq. INDEX RANGE SCAN     HR.PAY_ELEMENT_ENTRIES_F_N50     44           2
bq. bq. bq. bq. INDEX RANGE SCAN     HR.PAY_ELEMENT_ENTRY_VALUES_F_N50     9           3
And here is the plan when running inside pl/sql:
Operation Object Name Rows Bytes Cost Object Node In/Out PStart PStop
SELECT STATEMENT Optimizer Mode=CHOOSE 1 293463
bq. SORT AGGREGATE 1 48
bq. bq. MERGE JOIN CARTESIAN 1 M 56 M 293463
bq. bq. bq. MERGE JOIN CARTESIAN 78 3 K 3771
bq. bq. bq. bq. MERGE JOIN CARTESIAN 1 34 57
bq. bq. bq. bq. bq. TABLE ACCESS BY INDEX ROWID HR.PAY_ELEMENT_ENTRIES_F 1 17 29
bq. bq. bq. bq. bq. bq. INDEX RANGE SCAN HR.PAY_ELEMENT_ENTRIES_F_N50 44 3
bq. bq. bq. bq. BUFFER SORT 1 17 28
bq. bq. bq. bq. bq. TABLE ACCESS BY INDEX ROWID HR.PAY_ELEMENT_ENTRIES_F 1 17 28
bq. bq. bq. bq. bq. bq. INDEX RANGE SCAN HR.PAY_ELEMENT_ENTRIES_F_N50 44 2
bq. bq. bq. BUFFER SORT 15 K 107 K 3743
bq. bq. bq. bq. TABLE ACCESS BY INDEX ROWID HR.PAY_ELEMENT_ENTRY_VALUES_F 15 K 107 K 3714
bq. bq. bq. bq. bq. INDEX RANGE SCAN HR.PAY_ELEMENT_ENTRY_VALUES_F_N1 15 K 39
bq. bq. BUFFER SORT 15 K 107 K 289749
bq. bq. bq. TABLE ACCESS BY INDEX ROWID HR.PAY_ELEMENT_ENTRY_VALUES_F 15 K 107 K 3714
bq. bq. bq. bq. INDEX RANGE SCAN HR.PAY_ELEMENT_ENTRY_VALUES_F_N1 15 K 39
As you can see, the cost goes way up when running inside a pl/sql block.
Thank you very much for your help.
Eduardo Schñadower
Edited by: shinaco on Dec 12, 2008 4:10 PM
Added indentation. Sorry for that, I didn't realize copy-paste didn't work well here.

I meant to add,
Why is optimizer different in sql than in pl/sql? It is not different, it is the same optimizer.
If you replace bind variables with literals then you have a totally different query, with much more information available to the optimizer. If you are seeing the same statement with the same literal values get a different plan within PL/SQL than in SQL*Plus or TOAD etc then there must be something else different. Unfortunately there is a lot less diagnostic info available in 9i (and I don't have 9i around to test on). Perhaps you can get a 10053 trace from each session and see if you can see what is different.
btw I still can't read the execution plan. There is more information in the DBMS_XPLAN output, and it needs to be formatted using ** tags or equivalent HTML.

Similar Messages

WT SynchManager instl. on a different Server other than SAP/SQL Server.

Hi,
Is it possible to move/install WebTools SynchManager on a different server other than the SAP/SQL Server? Please let us know.
Note:
1. We are aware of the recommendation: u201CInstall the SynchManager on the SAP Business One server on which the DI API is installedu201D in the "Business One Web Tools SynchManager Guide.pdf".
2. But, SAP/SQL Server at our Client is starving of RAM due to a high number of windows services/application trying to run at the same time (although the maximum amount of RAM possible under the current configuration is installed (4GB)).
3. So, we are evaluating as to which of these application (WebTools being one of them) can be moved to a different server.
Thanks and Regards,
Kafil Shumon.

Hi Kafil,
You can put the synchmanager on a different server as long as the DI API is installed and that server can access the B1 licensing server.

Why does ical on icloud have a different lay out than on my mac

Why does iCal on my MacBook Pro have a different lay out than on iCloud? I like the iCloud layout better, the only missing niceties is the year calendar. How do I get both iCals to look the same? The ideal iCal? The one from iCloud WITH the year calendar.

Do a malware check with some malware scanning programs on the Windows computer. 
You need to scan with all programs because each program detects different malware. 
Make sure that you update each program to get the latest version of their databases before doing a scan. 
*http://www.malwarebytes.org/mbam.php - Malwarebytes' Anti-Malware
*http://www.superantispyware.com/ - SuperAntispyware
*http://www.microsoft.com/windows/products/winfamily/defender/default.mspx - Windows Defender: Home Page
*http://www.safer-networking.org/en/index.html - Spybot Search & Destroy
*http://www.lavasoft.com/products/ad_aware_free.php - Ad-Aware Free
See also:
*"Spyware on Windows": http://kb.mozillazine.org/Popups_not_blocked
Please update Firefox to the latest version via "Help > Check for Updates or Help > About Firefox" 
The Firefox version that you use is no longer maintained with security and stability updates. 
Using unsupported Firefox versions is at your own risk and makes you vulnerable to attacks. 
*Firefox 7.0.x: http://www.mozilla.com/en-US/firefox/all.html

Hi - for any photo, info in mountain lion shows a different file size than Photoshop does. I've tested with CS3 and CS6 and get the same problem. Doesn't happen with CS3 and OSX10.4 or any previous versions of Photoshop and OSX. Anyone know why?

Hi - for any photo, info in mountain lion shows a different file size than Photoshop does. I've tested with both CS3 and CS6 on 10.8.2 and get the same problem. This doesn't happen with CS3 and OSX10.4 or any previous versions of Photoshop and OSX. Anyone know why?

this one is actually a really rare symptom of a flaky connection to the ipod on a Windows PC. there's more going on in terms of hardware on nanos and 5th gens than in the earlier models ... so if the connection is flaky to precisely the right/wrong degree, itunes will see the ipod, but misidentify it as an earlier version of ipod.
tracking down the cause of the flakiness can be tricky ... as you already know ...
just checking. have you tried connecting with a different (known-good) USB cable? does that seem to have any impact on the rate of occurence of the problem?

Why do I keep getting this message - it is simply not true. 'You are signed in with a different Apple ID than previously used with Podcasts. Do you want to sync these podcast subscriptions and stations?

Why do I keep getting this message - it is simply not true. 'You are signed in with a different Apple ID than previously used with Podcasts. Do you want to sync these podcast subscriptions and stations?
<Email Edited by Host>

Please sign out of the iTunes Store by selecting
Store ▹ Sign Out
from the iTunes menu bar. Then sign back in.

Is Numbers different on iPad than on a Mac? Why don't I see toolbars with Numbers on my iPad?

Is Numbers different on iPad than on a Mac? Why don't I see toolbars with Numbers on my iPad?

So when I try to use a tutorial, it's gonna be a crapshoot if it will help me or not?

Why is my In design swatch panel creating swatches with different color builds than Pantone formulas

I am working in ID CS5 now but I have seen this problem in older versions as well. After creating a swatch, either process or spot, and using any of the Pantone selection choices, I get a different formula build than what is in the Pantone books. For example; PMS 110 PC is 2/24/100/7 in the book, but in InDesign it formulates as 0/12/100/7. Doers anyone have info on this?
S

maybe just as well, we are getting CS6 this month
CS6 includes Pantone+ Color Bridge libraries, which use CMYK Process definitions to simulate Pantone solid colors. The new solid library colors (i.e., Pantone+ Coated) are always defined as spot type with Lab definitions—you can no longer switch to CMYK definitions via Ink Manager for the those libraries.
Pantone+ Color Bridge Coated defines 110 as process CMYK 2|22|100|8. It's not surprising Pantone is always changing the CMYK simulation builds because the appearance of CMYK depends on the printing conditions. Personally I would go for a color managed conversion, allowing that much black in a saturated yellow seems questionable.

Different execution plans for the same sql

Hi,
Im testing our new 10gR2 database on Linux and I can't understand why the same query use different plan.
Here are the details.
Table name: invoice_detail
Records: About 10,670,900
Columns:
No
seq (the primary key is No+Seq). Each invoices contains +/- 10 invoice_details.
category ( <- 10 different values )
State ( <- 3 different values )
Basically, I have an index on the primary key and another index on Category + State.
My request:
select *
from invoice_detail
where no=123456 <- Best index to use
and state <> 'CANCEL'
and category = 'INVOICE'
If i run this query from Toad or sql+, that's fine.
The same query (i'm watching it from EM) executed via Forms use the category+state index.
When I first import the database, the last thing I do is to run DBMS_STATS.GATHER_DATABASE_STATS.
At this point, Forms use the right index.
The day after (after the database has been analyzed with the predefined job via EM) Forms use the wrong index.
I re-analyzed everything with exec DBMS_STATS.GATHER_DATABASE_STATS but the problem is still there.
Thanks in advance

I'm already using bind variables.
I changed the "Estimated Percentage" to 100% in "Gather Optimizer Statistics Default Options" and now it seems to use the correct index. I'm stressed because I dont understand why it chooses different plan for the same sql.
Actually, my users test the migration 1 day after I load all the data (drop schema-create schema-load data-analyze database) and at this point everythings go fine. After the second analyze of the database, the DB choose the wrong indexes.
I really cannot migrate until I understand why it happens.
Any ideas?
TIA

Why can´t i have more than one initiator task on my bpm process

Hi All.
I am working with BPM 11g and would like to know:
Why can´t i have more than one initiator task on my bpm process
In this case, i will have two separeted process where each has its initiator task?
Regards,
Diovani

Hi Diovani
Two different processes can have different task initiators...task initiator means creation of a process instance which can happen only once

Why is elements better at photomerge than CC- CC does not appear to automatically fill image based on content but elements does when merging a panorama. Also the stitching is visable in CC but almost perfect in elements- why?

I took 6 panorama shots of a scene and used CC to Photomerge them as one. Couldn't see where to automatic blend the edges and there was 'stitch' lines when the images were merged. So i did the same in Elements 11 and it was perfect. Am i doing something wrong in CC or perhaps not doing something at all?
Any help, please?
Dave

Hi - Thanks for taking the time to reply and i appreciate the remarks- if a little harsh- we all have to start somewhere and i am fully aware of the limitations of Elements which is why i decided to add CC to my software. I can only say that if an inferior quality software from Adobe does the job well then CC must also be suited to doing the same which is why i can only think, from your comments, that i have not done something simple- however- following tutorials to get to the end result should have sufficed- it didn't so perhaps i will consider posting the difference between the two applications- and, perhaps suffer a few more 'harsh' comments. The learning curve is quite steep and i am a visual learner, but i'm also not totally incompetent:)
Kind Regards
Dave Munn
Original message----
From : [email protected]
Date : 02/02/2015 - 06:45 (GMTST)
To : [email protected]
Subject : why is elements better at photomerge than CC- CC does not appear to automatically fill image based on content but elements does when merging a panorama. Also the stitching is visable in CC but almost perfect in elements- why?
why is elements better at photomerge than CC- CC does not appear to automatically fill image based on content but elements does when merging a panorama. Also the stitching is visable in CC but almost perfect in elements- why?
created by station_two in Photoshop General Discussion - View the full discussion
First a clarification: you are not addressing Adobe here in these user forums. You are requesting help from volunteers users just like you who give their time free of charge. No one has any obligation to answer your questions.
I'll give it my best shot anyway.
Few folks in this forum are really familiar with Elements, for which there's a dedicated, totally separate forum.
Different engineering teams, also.
From this perspective, it will be difficult to give you a direct answer to your "why?" question.
Personally, I blend very large panorama shots in Photoshop proper since I can't even remember when without any issues whatsoever, up to and including in Photoshop CS6 13.0.6.
Without being at your computer and without looking at your images, I couldn't even begin to speculate what you are doing wrong in Photoshop, which I suspect you are. The least you could show is post examples from your panoramas that have gone wrong.
I can tell you that panorama stitching requires significant overlap between the individual shots, besides common-sdense techniques like a very solid tripod and precision heads.
The only version of Elements I have ever used for any significant time was Elements 6 for Windows, which I bought in 2008 to use on a PC (I've been an avid Mac user for 30 years). I found Elements so limited and so bad that I successfully demanded a refund from Adobe. IU mention this only to emphasize that I can truly only address your question from a Photoshop (proper) and Mac user point of view. I couldn't care less about Elements, but if you have comparison examples of panoramas processed in both applications, by all means post those two.
Generally speaking Photoshop is a professional level application that makes no apologies for its very long and steep learning curve, while Photoshop has many hand-holding features for amateurs and beginners.
Perhaps the bottom line is that you should stick with Elements if you personally manage to get better results there.
If the reply above answers your question, please take a moment to mark this answer as correct by visiting: https://forums.adobe.com/message/7152397#7152397 and clicking ‘Correct’ below the answer
Replies to this message go to everyone subscribed to this thread, not directly to the person who posted the message. To post a reply, either reply to this email or visit the message page:
Please note that the Adobe Forums do not accept email attachments. If you want to embed an image in your message please visit the thread in the forum and click the camera icon: https://forums.adobe.com/message/7152397#7152397
To unsubscribe from this thread, please visit the message page at , click "Following" at the top right, & "Stop Following"
Start a new discussion in Photoshop General Discussion by email or at Adobe Community
For more information about maintaining your forum email notifications please go to https://forums.adobe.com/thread/1516624.

Why SCCM 20012 Install fails with remote 64 bit SQL Server but proceeds for remote 32 bit SQL Server

I have the following setup:
AD User accounts
- SQLAdmin (used to run SQL Services i.e. Database Engine)
- SCCMADmin (used to install SCCM 2012)
SPN registered for user account SQLAdmin
setspn -S MSSQLSvc/SQL.Domain_Name:1432 Domain_Naem\SQLAdmin
setspn -S MSSQLSvc/SQL:1432 Domain_Name\SQLAdmin
setspn -S MSSQLSvc/SQL.Domain_Name:1433 Domain_Naem\SQLAdmin
setspn -S MSSQLSvc/SQL:1433 Domain_Name\SQLAdmin
and checked with setspn -L TESTING\SQLAdmin
1 SERVER called SQL
32bit Server with SQL Server 2008 + SP3 + Cumulative Update 6
2 SQL instances - Default (MSSQLSERVER) and
SCCM2012
MSSQLSERVER instance uses TCP port 1433
SCCM2012 Instance users TCP port 1432 (no dynamic ports)
User accounts that have been given public and sysadmin SQL server roles
on both instances are: SQLAdmin, SCCMAdmin, Domain Administrator, Local Administrator and computer account SCCM
Client Protocols TCP enabled (and Named Pipes) (checked via SQL Server Configuration Manager)
Local Administrators Group on this server has members - SQLAdmin, SCCMAdmin, Domain Administrator, Local Administrator and computer account SCCM
Firewall turned on with access allowed on Ports 1432, 1433, 4022,445, and WMI - WMI-in, DCOM-in and ASync-iN builtin rules allowed\enabled.
1 SERVER called SQL3
64bit Server with SQL Server 2008 + SP3 + Cumulative Update 6
2 SQL instances - Default (MSSQLSERVER) and
SCCM2012
MSSQLSERVER instance uses TCP port 1433
SCCM2012 Instance users TCP port 1432 (no dymanic ports)
User accounts that have been given public and sysafmin SQL server roles on both instances are:
SQLAdmin, SCCMAdmin, Domain Administrator, Local Administrator and computer account SCCM
Client Protocols TCP enabled (and Named Pipes) (checked via SQL Server Configuration Manager)
Local Administrators Group on this server has members - SQLAdmin, SCCMAdmin, Domain Administrator, Local Administrator and computer account SCCM
Firewall turned off
1 SERVER called SCCM
64bit Server that is to be the Primary Site Server\MP for SCCM 2012
ODBC link to SQL\SCCM2012,1432
ODBC link to SQL3\SCCM2012,1432
ODBC uses SQL Native Client 10.0 (64 bit)
Both ODBC connections when TESTed pass and suggest connectivity to SQL Servers
Install process doe SCCM2012
Tried to install SCCM 2012 RC2 when logged in to SCCM Server as AD user account
SCCMAdmin, and when utilising the SCCM2012 SQL instance on
32 bit server of SQL install proceeds barring warning about 8GB rec, for SQL Server. Then fails on PKI certificate issue. Installation (chose HTTP for MP). I beleive the PKI failure as install starts is
due to the fact that SCCM 2012 needs its database server to b 64 bit ?
Thus i then tried to install SCCM 2012 RC2 when logged in to SCCM Server as AD user account SCCMAdmin, and
when using the SCCM2012 SQL instance on
64 bit server of SQL install proceeds but fails at checking stage and says:
SQL Server sysadmin rights FAILED SQL3.Domain_Name
Either the user account running Configuration Manager Setup does not have sysadmin SQL Server role permissions on the SQL Server instance selected for site database installation, or the SQL Server instance could not be contacted to verify permissions. Setup
cannot continue.
and
Site System to SQL Server communication WARNING SCCM.Domain_Name
A communication error has been detected between the specified site system and the site database computer. This error can occur when the site database server is offline or if a valid SPN has not been registered in Active Directory Domain Services for the SQL
Server instance hosting the site database. Setup cannot continue.
Why does the install of SCCM 2012 with 32 bit SQL proceed further than the install with 64 bit SQL, with the latter process failing as above error meesages show and yet both servers are set identically (apart from temporary turning off Firewall on the 64
bit server) and during the install the Databse Server specified is accepted ?

Thanks for the reply.
I can connect via ODBC to the the 32 bit SQL Server and the 64 bit SQL Server from the SCCM Server
The SQL Server Unit called SQL3 unit is a 64 bit SQL.
The SQL Server Unit called SQL is only 32 bit, but at least gets past the final checking stage and the error messages about sysadmin rights and Site System to SQL Server communication problem, and then fails with PKI certificate error message.
When trying to install SCCM specifying the 64 bit SQL Server as the Database Server it gives the 2 error messages at the final checking stage of the installation as listed.
So as I said what is confusing is the fact that if the SCCM install uses the remote 32 bit SQL Server it passes the final checking (although fails with PKI certificate message) but the 64 bit SQL Server set up exactly the same apart from the Firewall being
left off for the time being, fails at the final check stage with the 2 listed errors.
If I use a local 64 bit SQL Server the installation is fine.
Still would like to find out what cause the 2 issues for the remote 64 bit SQL Server, when ODBC seems fine, sysadmin rights have been given for the installer account and the SCCM computer account and SPNs have been set for the user account running the SQL
Services.

Why is audio file size larger than in Flash?

I use both Flash and Captivate. Final file size is a big consideration. I wondered why the file size of the audio in Captivate is almost twice as large as the same audio if it's used in Flash. I'm using the same audio settings for both. Does anyone know what causes the audio compression to be different in Captivate than in Flash?

OK, sounds like memory will be fine, so let's look at some other stuff...
Is it only one particularly keynote presentation that slows up for you, or does keynote behave in the same way even if you start from scratch and create a new presentation?
If it's just the one presentation then there's probably some content (possibly video) within it that KN takes exception to. Sometimes a single problematic slide or piece of content can slow down the whole app.
Is the problematic presentation one that was imported from Powerpoint? If so, import it again and look carefully at the import error/warning logs.
Before you start keynote, start up activity monitor - which will give you an indication of what programs are using all the CPU/memory/disk bandwidth etc. You'll then have an idea of whether it is KN maxing out on CPU or whether anything else is dragging the system down.
Another thing you could try would be to log in to a different user account on your machine (create one if necessary) and then try opening the presentation from within this different account. If this works for you then it suggests a problem with the keynote plist file in your original user account.
I don't know if it'd be acceptable to you, but if you could post a link to your problematic presentation then I could download it and try it out on my normally reliable Keynote on one of my computers. (Make sure that you've ticked the box to embed all media in the presentation)
Hope some of this is helpful,
Mike.

Why the flashback log'size smaller than the archived log ?

hi, all . why the flashback log'size smaller than the archived log ?

Lonion wrote:
hi, all . why the flashback log'size smaller than the archived log ?Both are different.
Flash logs size depends on parameter DB_FLASHBACK_RETENTION_TARGET , how much you want to keep.
Archive log files is dumped file of Online redo log files, It can be either size of Online redo log file size or less depending on online redo size when switch occurred.
Some more information:-
Flashback log files can be created only under the Flash Recovery Area (that must be configured before enabling the Flashback Database functionality). RVWR creates flashback log files into a directory named “FLASHBACK” under FRA. The size of every generated flashback log file is again under Oracle’s control. According to current Oracle environment – during normal database activity flashback log files have size of 8200192 bytes. It is very close value to the current redo log buffer size. The size of a generated flashback log file can differs during shutdown and startup database activities. Flashback log file sizes can differ during high intensive write activity as well.
Source:- http://dba-blog.blogspot.in/2006/05/flashback-database-feature.html
Edited by: CKPT on Jun 14, 2012 7:34 PM

Iphone6 why can't I save more than one contact?

iphone 6 why can't i save more than one contact?

Hi Diovani
Two different processes can have different task initiators...task initiator means creation of a process instance which can happen only once

You are signed in with a different Apple ID than previously used with Podcasts. Do you want to sync these podcast subscriptions and stations with "******"?

Why am I getting this error message every time I open iTunes? I'm giving the option of Sync or don't Sync and click sync and there are no problems. But why does it open with: You are signed in with a different Apple ID than previously used with Podcasts. Do you want to sync these podcast subscriptions and stations with “*****”?
<Edited by Host>

Please sign out of the iTunes Store by selecting
Store ▹ Sign Out
from the iTunes menu bar. Then sign back in.

Why is optimizer different in sql than in pl/sql?

Similar Messages

Maybe you are looking for