Select distinct performance

Hi,
I have the following issue. A table with appox 650,000 rows and 200 columns is being queried for distinct values on a column(SN) that is of type VARCHAR2(200 Bytes).
The query returns 40 distinct values in 30 secs. I would like to bring that time down.
The original DBA had created a single index on two columns (SN, DN). Column DN is of type NUMBER(10,0).
When I run the query with autotrace on I see that the index is not being used and a full table scan is being done.
I tried creating an index on just SN, that is not being used either.
Any suggestions?
Thanks.

2) I'm confused about timings. You said initially that it took 30 seconds to retrieve the 40 rows. But your autotrace output seems to indicate that it took 1 second and only did 657 consistent gets. What am I missing?
--Hmm. could be because I used the 'explain plan for' command. Here's the entire run:*
41 rows selected.
Elapsed: 00:00:36.48
Execution Plan
   0      SELECT STATEMENT Optimizer=ALL_ROWS (Cost=31916 Card=37 Byte
          s=555)
   1    0   SORT (UNIQUE) (Cost=31869 Card=37 Bytes=555)
   2    1     TABLE ACCESS (FULL) OF 'TABLE' (TABLE) (Cost=2
          9817 Card=627746 Bytes=9416190)
Statistics
          1 recursive calls
          0 db block gets
      68943 consistent gets
      68810 physical reads
          0 redo size
       1415 bytes sent via SQL*Net to client
        534 bytes received via SQL*Net from client
          4 SQL*Net roundtrips to/from client
          1 sorts (memory)
          0 sorts (disk)
         41 rows processed3) Are you certain it is appropriate to run this query in the first place? I would be sorely tempted to believe that there ought to be a second table that had just 40 rows of data and that the table you're currently looking at should have a foreign key relationship to that other table. Your query could then simply ask for the 40 rows of data from the 40 row table.
--Ahh! But there in lies the problem. The previous DBA who put this together was unfortunately not a DBA. This person was pulled in from manufacturing and told to learn what they could about oracle and put together a database. Now, that that person's gone I have to deal with it. A whole app was built on top of this db and I cannot change its design without ticking off a lot of people. So, I am stuck. *

Similar Messages

Serious performance problem - SELECT DISTINCT x.JDOCLASSX FROM x

I am noticing a huge performance problem when trying to access a member that
is lazily loaded:
MonitorStatus previousStatus = m.getStatus();
This causes the following query to be executed:
SELECT DISTINCT MONITORSTATUSX.JDOCLASSX FROM MONITORSTATUSX
This table has 3 million records and this SQL statement takes 3 minutes to
execute! Even worse, my app heavily uses threads, so this statement is
executed in each of the 32 threads. As a result the application stops.
Is there any way that I can optimize this? And more importantly, can Kodo
handle a multithreaded app like this with a huge database? I've been having
a lot of performance problems since I've started doing stress & load
testing, and I'm thinking Kodo isn't ready for this type of application.
Thanks,
Michael

You can prevent this from happening by explicitly enumerating the valid
persistent types in a property. See
http://docs.solarmetric.com/manual.html#com.solarmetric.kodo.PersistentTypes
for details.
>
Inconveniently, this nugget of performance info is not listed in the
optimization guide. I'll add in an entry for it.This setting did in fact prevent the query from running which fixed the
problem. It definitely belongs in the optimization guide.
And more importantly, can Kodo
handle a multithreaded app like this with a huge database? I've beenhaving
a lot of performance problems since I've started doing stress & load
testing, and I'm thinking Kodo isn't ready for this type of application.I'd like to find out more information about details about your issues. We
do a decent amount of stress / load testing internally, but there are
always use cases that we don't test. Please send me an email (I'm assuming
that [email protected] is not really your address) and let's
figure out some way to do an analysis of what you're seeing.This email is just for posting to usenet, to avoid spam. I'm now running my
app through stress/load testing so I hope to discover any remaining issues
before going into production. As of this morning the system seems to be
performing quite well. Now the biggest performance problem for me is the
lack of what I think is called "outer join". I know you'll have this in 3.0
but I'm suprised you don't have this already because not having it really
affects performance. I already had to code one query by hand with JDBC due
to this. It was taking 15+ minutes with Kodo and with my JDBC version it
only takes a few seconds. There are lots of anti-JDO people and performance
issues like this really give them ammunition. Overall I just have the
impression that Kodo hasn't been used on many really large scale projects
with databases that have millions of records.
Thanks for configuration fix,
Michael

Issue with "Select Distinct" query in Oracle 10g against Oracle 9i

Hi,
I would appreciate if some one help me here because it is really urgent.
We are upgrading our database from 9i to 10g.
There are the "Select distinct" queries in the code which populated the grid on the applications screens. We found a difference in 9i and 10g the way the result is populated for these queries. If "Select Distinct" query wihtout a order by clause is executed in 9i then the result is automatically sorted. But Oracle 10g does not do this.
We can change the queries adding order by clause but we are almost at the end of the testing and want to know if there is any way that we can do this from database settings. Would there be any impact of these settings change on overall operation of Oracle 10g?
I would appreciate if some one can help me here.
Thanks,
Dinesh

then the result is automatically sorted.No. Oracle may have done a sort operation to perform the distinct, but it still did not guarantee the order of your results.
In 10g and in 9i, if you want your results in a certain order you must use order by.

Select Distinct and join in ODI

Hi,
I have following task to perform: I am loading metadata into Planning dimension from Oracle database. I have two tables
1. "Sales"
Columns: Name, Number, Value
Sample Data:
Product 1, 10, 200
Product 2, 30, 100,
Product 1, 15, 500
2. P&R
Columns:
Name, Alias
Product 1, SampleSoda1
Product 2, SampleSoda2,
Resource 1, CanForSoda,
Resource 2, CO2
What I need to do is: I have to select name and alias from second table of all products that were sold.
So I need to select distinct Name from Table 1, naxt join it with Table 2 (so I have Name and Alias) and load it to planning.
I am a little confused how to do it.
Any help would be great!
Best regards,
Greg

Hi Greg,
What you can do is either :
- Create a yellow interface with your table Sales as source. Map the name column directly in the target. In the flow tab, click on your target and select "Distinct rows".
- Create a second interface, with your first interface as source. Select the "Use Temporary Interface as Derived Table (Sub-Select)" checkbox.
- Add your second datastore and join it. Or you can use a lookup table.
OR
- Create an interface with Sales and P&R as source (or set P&R as a lookup table).
- Go on the flow tab and select "Distinct rows".
If you've a lot a data in the first table, I would go for the first solution.
Hope it helps.
Regards,
JeromeFr
Edited by: JeromeFr on Feb 14, 2013 9:52 AM
To be more clear in the first step of solution 1

SELECT DISTINCT With OPEN cursor FOR

Hello.
I have the following procedure. All it does is open a cursor for an SQL string passed into it, and return the open cursor.
PROCEDURE sp_execute_dynamic (hold_input_string IN CLOB,
hold_cursor OUT hold_cursor_type) IS
BEGIN
OPEN hold_cursor FOR TO_CHAR(hold_input_string);
END sp_execute_dynamic;
It works fine except when I perform SELECT DISTINCT. I get the following error.
SQL> declare
2 TYPE hold_cursor_type IS REF CURSOR;
3 hold_cursor hold_cursor_type;
4 hold_object_name VARCHAR2(1024);
5 hold_object_type VARCHAR2(1024);
6 begin
7 dynamic_sql_pkg.sp_execute_dynamic('select distinct object_name from user_objects where object_
name in (''PLAN_TABLE'',''DBA_OBJECTS'')',hold_cursor);
8 loop
9 fetch hold_cursor into hold_object_name, hold_object_type;
10 exit when hold_cursor%NOTFOUND;
11 dbms_output.put_line('Object Name = '||hold_object_name||' Object Type = '||hold_object_type);
12 end loop;
13 end;
14 /
declare
ERROR at line 1:
ORA-01007: variable not in select list
ORA-06512: at line 9
It does the same thing with SELECT UNIQUE or SELECT with a GROUP BY. Can anyone tell me why this happens and what I could to to work around it?
Thanks
Chris

see at line 7 you are selecting only one column and at line 9you are fetching into two variables
7 dynamic_sql_pkg.sp_execute_dynamic('select distinct object_name from user_objects where object_
name in (''PLAN_TABLE'',''DBA_OBJECTS'')',hold_cursor);
8 loop
9 fetch hold_cursor into hold_object_name, hold_object_type;
HTH

Select distinct from an infoset query

Hello
I need to select distinct / delete duplicate from an infoset query created thru SQ02.
Please let me to know how this can be done?
Many thanks in advance!
regards
Sanjyot

Hi,
You can verywell use infoset query if you are doing following things
Join using Keyfields to retreive data from tables
If not using keyfields create index for those table fields in those fields which will improve in accessing database
Try to use minimum of tables of small size dont try to join big tables like GLPCA and all.
Try to load small set of data like a period or month. Dont try for a year or so.
if your performance is good for aperiod you can go for a year.
If you are good in function module you can try function module in which you will have the option of specifying no of records to be selected usign package size.
Hope this helps for you.
Thanks,
Arun

In (select distinct .... ) Vs exists

Guys, which would be less costly and also quicker performance:
SELECT a.*
          FROM table1 a
WHERE a.linkid IN (
          SELECT DISTINCT b.id1
                     FROM table2 b
                    WHERE b.REF = 'SOMETHING'
          UNION ALL
          SELECT DISTINCT b.id2
                     FROM table2 b
                    WHERE b.REF = 'SOMETHING'
          UNION ALL
          SELECT DISTINCT b.id3
                     FROM table2 b
                    WHERE b.REF = 'SOMETHING'
          UNION ALL
          SELECT DISTINCT b.id4
                     FROM table2 b
                    WHERE b.REF = 'SOMETHING'
          UNION ALL
          SELECT DISTINCT b.id5
                     FROM table2 b
                    WHERE b.REF = 'SOMETHING'
          UNION ALL
          SELECT DISTINCT b.id5
                     FROM table2 b
                    WHERE b.REF = 'SOMETHING'
          UNION ALL
          SELECT DISTINCT b.id7
                     FROM table2 b
                    WHERE b.REF = 'SOMETHING')
SELECT a.*
FROM table1 a
WHERE EXISTS (
          SELECT 1
            FROM table2 b
           WHERE b.REF = 'SOMETHING'
             AND (   a.linkid = b.id1
                  OR a.linkid = b.id2
                  OR a.linkid = b.id3
                  OR a.linkid = b.id4
                  OR a.linkid = b.id5
                  OR a.linkid = b.id6
                  OR a.linkid = b.id7)
                   )I have a feeling the optimiser may execute both with the same plan.
Thanks, FIXXXER
Edited by: fixxxer on 04-Nov-2011 05:54

Depends on a lot of things, one of them being your version.
It's possible Oracle would rewrite your query using a myriad of techniques, one of the newer ones being an OR EXPANSION
http://blogs.oracle.com/optimizer/entry/or_expansion_transformation
One thing to note is that Oracle knows not to add the DISTINCT's inside of the IN clause .... because when you say IN (some_set) that set is always DISTINCT'd for you.

Select distinct and 2 inner joins in same select statement

can anyone guide me how to write the below query for higher performance. can I use for all entires or shall i split this,
please let me know your opinions.
SELECT DISTINCT werks matnr b~stlal idnrk postp d~aennr
   sortf potx1 c~lkenz b~stlty stlst
   INTO TABLE ltab_one
   FROM mast AS a INNER JOIN stko AS b
     ON a~stlnr = b~stlnr
     AND a~stlal = b~stlal
     INNER JOIN stas AS c
     ON b~stlnr = c~stlnr
     AND b~stlal = c~stlal
     AND b~stlty = c~stlty
     INNER JOIN stpo AS d
     ON c~stlnr = d~stlnr
     AND c~stlkn = d~stlkn
     AND c~stlty = d~stlty
WHERE a~werks IN s_werks
     AND ( ( d~andat IN s_datum
        OR d~aedat IN s_datum
        OR a~andat IN s_datum
        OR a~aedat IN s_datum
        OR b~andat IN s_datum
        OR b~aedat IN s_datum
        OR c~andat IN s_datum
        OR c~aedat IN s_datum
        ) AND d~datuv >= lv_effdt )
     AND a~stlan = '1'
     AND matnr IN s_matnr.

WHERE a~werks IN s_werks
     AND ( ( d~andat IN s_datum
        OR d~aedat IN s_datum
        OR a~andat IN s_datum
        OR a~aedat IN s_datum
        OR b~andat IN s_datum
        OR b~aedat IN s_datum
        OR c~andat IN s_datum
        OR c~aedat IN s_datum
        ) AND d~datuv >= lv_effdt )
     AND a~stlan = '1'
     AND matnr IN s_matnr.
This part determines your performance, not the DISTINCT and not the joins.
What is in s_werks, s-datum, s_matnr typically? I guess there are better and worse conditions.

How to undo "select distinct" in OBIEE

Hello guys
I found that pretty much in all the reports that we created in OBIEE, the presentation service would generate SQL that starts with "select distinct" when I view the query log..
I'd like to know what configuration in the Admin Tool or presentation service is controlling this behavior of OBIEE. For some reason as I am tuning the performance by running the same query in DB, the same query without "distinct" can go so much faster than the original one from OBIEE.. So I'd like to configure my RPD so that certain reports will not have "select distinct". I know that by default, the BI server will always generate a "distinct", but is there a way to override this default, if not, what might be some performance tuning tactics I can use?
Any thoughts?
Thanks
Edited by: user7276913 on Oct 27, 2009 8:20 PM

BenS wrote:
Two issues with this that I encountere:
When I selected the first suggested option at the Physical Layer for the database, I found the "DISTINCT SUPPORTED" option. There were two columns checked: Value and Default. I was able to unselect Value but not Default.
(1) What is the difference between Value and Default?
(2) If I only unselect Value will it stop Answers from adding the DISTINCT to every query created?
Second, when I selected the source from the business model mapping, it already did not have the DISTINCT box checked. Therefore, I could not limit this at the Logical layer as I would have preferred.
If you could help me understand the first option it may clarify how to implement option # 2.
Thanks...1) the defaults are stored in an .ini file and serve the purpose of defining the defaults when you hit the 'revert to default' button.
If you unselect value, you are telling OBIEE that the database for which you are changing these options does not support the distinct clause, I doubt you wish to do this across the entire application.
I think you will find that OBIEE applies the distinct :
1) when the LTS has distinct ticked.
2) When the column in question is not part of a dimension hierachy key (if it is, your telling the OBIEE server these should be unique, so wont issue the distinct)
or
3) the column forms part of a key in either logical BMM layer or as part of a physical key.
adding the column to a Dim Hierachy wont be an option in most cases, you could try playing with just setting this column as a key on the BMM layer or Physical layer , and let us know if this helps.

SELECT DISTINCT with LINQ

Hey,
I am trying to perform a SELECT DISTINCT COL1, COL2, COL3 FROM VIEW WHERE CONDITION using LINQ with ODAC 112030 but I get different results.
C#:
var s1 = context.Select(v=>new { v.COL1, v.COL2, v.COL3}).Where(...).Distinct();
Any ideas?

Hey,
I think this way works - with anonymous types:
context.Entity.Where(...).Select(ent => new { col1 = ent.COL1, ... }).Distinct();.
You can also do context.Entity.Select("distinct it.COL1, it.COL2").Where("..."); but one should be careful to add all necessary columns in select statement.
Edited by: 917334 on Jun 7, 2012 7:11 AM

Comparing SELECTs for performance

I have a long-running function module and through SAT, I've identified its SELECT from BSEG as a potential hotspot and way that I can improve the run time, etc.
Through some research in SCN, I found I can/should use one of the component tables (BSID) to improve performance, but I don't know *how much* of an improvement it will be.
Rather than change the code of the FM and transport it to QA just to analyze performance gains, I was hoping there was some sort of tool that I could use to analyze separate code comparatively.
I thought of making 2 Quickviewer queries (one for BSEG and one for BSID) and then using ST05 to analyze each. Is there a better way?
Basically in PRD, I'd like to compare isolated SELECTs for performance to see what gains I could make.
Thanks
Jeremy H.

You can find some interesting discussions linked here, some also talk about index access:
FAQ's, intros and memorable discussions in the ABAP Testing and Troubleshooting Space
Regarding your point 4), make sure your WHERE-condition contains as many fields of the index top down without gaps.
Looking at BSID~4:
MANDT
Client
BUKRS
Company Code
REBZG
Number of the Invoice the Transaction Belongs to
REBZJ
Fiscal Year of the Relevant Invoice (for Credit Memo)
REBZZ
Line Item in the Relevant Invoice
KUNNR
Customer Number
UMSKS
Special G/L Transaction Type
REBZT
Follow-On Document Type
MANDT will be included automatically. BUKRS is not very selective (not many distinct values), but you must include it since it is the leading field of the index. REBZG sounds like a selective field, so you can likely use the index effectively by having just BUKRS and REBZG in your WHERE-condition, but the more the better.
Also have a look at transactions DB05 and TAANA which can give you a good idea about the actual data distribution.
Thomas

Toplink issues select distinct type query before fetch

When we have type classifiers identified by value, Toplink issues a 'SELECT DISTINCT TYPE' on the parent object and then queries the child records for that type, is there way to avoid the select distinct
type query from Toplink in queries, this is slowing down the query performance. thanks

More info
This is the Report Query:
Session session = PersistenceFactory.newSession();
ExpressionBuilder eb = new ExpressionBuilder();
ReportQuery query = new ReportQuery( eb );
query.setReferenceClass( FXSingleLegDealC.class );
query.useCollectionClass( ArrayList.class );
query.setSelectionCriteria( eb.get( "objectID" ).equal( 249994 ) );
query.addAttribute( "objectID", eb.get( "objectID" ) );
query.addAttribute( "parentDealFXDealLegCurrency1Amount", eb.get( "dealFXSingleLegParent" ).get( "fxDealLeg" ).get( "currency1Amount" ) );
query.addAttribute( "parentDealFXDealLegCurrency1Amount", eb.get( "dealFXSingleLegParent" ).get( "fxDealLeg" ).get( "currency2Amount" ) );
query.addAttribute( "parentDealFXDealLegCurrency1Amount", eb.get( "dealFXSingleLegParent" ).get( "fxDealLeg" ).get( "currency1" ).get( NamedEntity.ShortName) );
query.addAttribute( "parentDealFXDealLegCurrency1Amount", eb.get( "dealFXSingleLegParent" ).get( "counterpartyA" ).get( NamedEntity.ShortName) );
ArrayList result = ( ArrayList ) session.executeQuery( query );
This is the SQL Generated:
SELECT t0.id, t1.ccy1Amt, t1.ccy2Amt, t2.shortName, t3.shortName FROM IdcCpty t3, IdcInstr t2, IdcDeal t1, IdcDeal t0
WHERE (((t0.id = ?) AND (t0.type = ?)) AND ((((t1.id = t0.parentId) AND (t1.type = ?)) AND ((t2.id = t1.ccy1Id) AND (t2.type = ?))) AND ((t3.id = t1.cptyAId) AND (t3.type = ?))))
bind => [249994, IdcFXSLegDealC, IdcFXSLegDealC, IdcCrncC, IdcLglEntC]
Not sure why we need to query for type on t2 and t3 after we have made a join by id (t2.id = t1.ccy1Id,t3.id = t1.cptyAId), this is really breaking our performance, we do not want to add an index on
the type column
can someone please respond!?

Query help in select Distinct on one column.

CREATE GLOBAL TEMPORARY TABLE Table1 (
ID1 varchar2(100) ,
Name1 varchar2(100),
Name11 varchar2(100)
insert into Table1 values ('a','n1','h3');
insert into Table1 values('b','n2','h2');
insert into Table1 values('a','n3','h1');
insert into Table1 values('c','n4','h5');
insert into Table1 values ('c','n5','h4');
insert into Table1 values('d','n6','h6');
select * from Table1;
ID1,NAME1,     NAME11
a,     n1,     h3
b,     n2,     h2
a,     n3,     h1
c,     n4,     h5
c,     n5,     h4
d,     n6,     h6
I am trying to select distinct ID1 and all values associated with it which is max row.I want to result as -
ID1,NAME1,     NAME11
a,     n3,     h1
b,     n2,     h2
c,     n5,     h4
d,     n6,     h6
Can you please help me to write simple query to get above result.
Edited by: 871447 on Jul 25, 2011 9:42 AM
Edited by: 871447 on Jul 25, 2011 9:45 AM

Hi,
Do a self-join, to combine the two rows for each value of id1 onto one output row.
Make it an outer join, in case there is only one row with a vlaue for id1.
SELECT l.id1
,     l.name1
,     NVL ( r.name11
         , l.name11
         )          AS name11
FROM              table1     l
LEFT OUTER JOIN     table1     r ON l.id1     = r.id1
                          AND l.name1     < r.name1
;Edited by: Frank Kulash on Jul 25, 2011 12:57 PM
Sorry, I mis-read the problem.
Lee's solution, above, assumes that name1 is unique, as it is in your sample data.
What output would you want if that's not the case?
If name1 is not unique, but the combination of (id1, name1) is unique, then you can modify Lee's solution like this:
SELECT *
FROM    table1
WHERE   (id1, name1) IN (
                    SELECT    id1
                    ,       MAX (name1)
                           FROM          table1
                    GROUP BY id1
;Or, if you can't make any assumptions about uniqueness, you might need something like this:
WITH     got_r_num     AS
     SELECT id1, name1, name11
     ,     ROW_NUMBER () OVER ( PARTITION BY id1
                               ORDER BY         name1     DESC
                         ,                name11     DESC
                       )      AS r_num
     FROM     table1
SELECT id1, name1, name11
FROM     got_r_num
WHERE     r_num     = 1
;

Select List Performance Issue in Apex 3.1.2

Hi
It would of great help if you could suggest us for the following
We are facing some issues in Select List performance in Apex 3.1.2 version.
We have 6 select list and all have been cascaded i.e., values of each select list based on pervious list value.
Values in each select list are huge. Because of this, the performance is very slow.
It takes huge time to fetch the data based on other list values.
We cross verified in backend, the same query takes less time compared to the Application.
Any recommedation to fine tune this?
Thanks in advance
Vijay

if your select lists are very huge then it could be the browser that is causing the slow down.
try and create an example using a static HTML file containing a select list of the same size and you may find the same performance issue.
a quick way to create the test would be to save the source of your APEX page as a html file
Craig
[http://www.oracleapplicationexpress.com]

How to select distinct values from a table when it has composite primary ke

Hi
I have the requirement like , I need to select distinct one column values from the table which has composite primary key. How to acheive this functioinality using view object.
Eg : Table 1 has col1 and col2, col3
col1 col2 col3
1 A NA
1 B NA
2 A NA
3 C NA
2 D NA
primary key (col1,col2)
I have to select distinct col1.
Thanks

Hi
I got the solution for above. By Creating the read only view object we can acheive this.
thanks

Select distinct performance

Similar Messages

Maybe you are looking for