Query using large(in GB) temp tablespace, how to tune it?

Hi,
Below query is using large Temp tablespace, please guide for tuning this query so that it can use minimal temp space.
SELECT DISTINCT cust_alt.cust_acct_alt_id, cust_alt.cust_acct_alt_id_type_cd,
                trans.legal_entity_id, trans.exchange_id,
                trans.stamp_update_dtime, trans.book_alt_id,
                trans_alt.trans_alt_id, trans_alt.trans_alt_vers_id,
                trans_alt.src_sys_short_name, event.trans_event_sequence_id,
                event.trans_event_type_cd, prod_alt.product_alt_id,
                prod_alt.product_alt_id_type_cd, trans.buy_sell_cd,
                trans.trade_date, item.setlmt_date, item.trans_price_amt,
                item.price_yield_ind, item.trans_dirty_price_amt,
                item.trans_qty, item.trans_item_type_cd,
                event.trans_event_dtime,
                DECODE (extnn.confirm_generation_elig_ind,
                        'Y', 'In Progress',
                        NULL, 'Not Processed',
                        'Confirm Ineligible'
                       ) confirm_status,
                event_ref.trans_event_type_desc
           FROM ods_trans_event_type_ref event_ref,
                ods_trans_type_ref trans_type,
                ods_cust_acct_alt_id cust_alt,
                ods_product_alt_id prod_alt,
                ods_trans_item item,
                ods_trans_alt_id trans_alt,
                ods_trans trans,
                ods_trans_extnn extnn,
                ods_trans_event event
          WHERE event_ref.trans_event_type_cd = event.trans_event_type_cd
            AND trans_type.trans_type_id = trans.trans_type_id
            AND prod_alt.product_id = trans.product_id
            AND cust_alt.cust_acct_id = trans.cust_acct_id
            AND item.trans_vers_id = trans.trans_vers_id
            AND item.trans_id = trans.trans_id
            AND extnn.trans_vers_id(+) = trans.trans_vers_id
            AND extnn.trans_id(+) = trans.trans_id
            AND trans_alt.trans_link_reason_cd = 'FOID'
            AND trans_alt.trans_vers_id = trans.trans_vers_id
            AND trans_alt.trans_id = trans.trans_id
            AND trans.trans_vers_id = event.trans_vers_id
            AND trans.trans_id = event.trans_id
            AND item.trans_item_type_cd IN ('MAIN', '1', '2')
            AND cust_alt.cust_acct_alt_id != 'BOOKTOBOOK'
         order by trans_alt.trans_alt_id, event.trans_event_sequence_idThe Explain for this query is
| Id | Operation                                      | Name                        | Rows | Bytes | Cost (%CPU)| Time     |
|   0 | SELECT STATEMENT                          |                             |    79 | 23384 | 1404   (1)| 00:00:17 |
|   1 | HASH UNIQUE                                   |                             |    79 | 23384 | 1404   (1)| 00:00:17 |
|* 2 |   TABLE ACCESS BY INDEX ROWID        | ODS_TRANS_ITEM       |     1 |    53 |     4   (0)| 00:00:01 |
|   3 |    NESTED LOOPS                                |                             |    79 | 23384 | 1403   (1)| 00:00:17 |
|   4 |     NESTED LOOPS OUTER                    |                             |    73 | 17739 | 1144   (1)| 00:00:14 |
|   5 |      NESTED LOOPS                              |                             |    73 | 16717 |   965   (0)| 00:00:12 |
|   6 |       NESTED LOOPS                             |                             |    73 | 15476 |   819   (0)| 00:00:10 |
|   7 |        NESTED LOOPS                            |                             |    73 | 14089 |   673   (0)| 00:00:09 |
|   8 |         NESTED LOOPS                           |                             |    73 | 10439 |   672   (0)| 00:00:09 |
|   9 |          NESTED LOOPS                          |                             |    90 | 9810 |   312   (0)| 00:00:04 |
|* 10 |           TABLE ACCESS BY INDEX ROWID| ODS_TRANS_ALT_ID            |    92 | 3680 |    52   (0)| 00:00:01 |
|* 11 |            INDEX RANGE SCAN               | ODS_TRANS_ALT_ID_X1         |   119 |       |     5   (0)| 00:00:01 |
| 12 |           TABLE ACCESS BY INDEX ROWID| ODS_TRANS                   |     1 |    69 |     3   (0)| 00:00:01 |
|* 13 |            INDEX UNIQUE SCAN              | ODS_TRANS_PK                |     1 |       |     2   (0)| 00:00:01 |
| 14 |          TABLE ACCESS BY INDEX ROWID | ODS_TRANS_EVENT             |     1 |    34 |     4   (0)| 00:00:01 |
|* 15 |           INDEX RANGE SCAN                | ODS_TRANS_EVENT_PK          |     1 |       |     3   (0)| 00:00:01 |
| 16 |         TABLE ACCESS BY INDEX ROWID | ODS_TRANS_EVENT_TYPE_REF    |     1 |    50 |     1   (0)| 00:00:01 |
|* 17 |          INDEX UNIQUE SCAN                | ODS_TRANS_EVENT_TYPE_REF_PK |     1 |       |     0   (0)| 00:00:01 |
| 18 |        TABLE ACCESS BY INDEX ROWID   | ODS_PRODUCT_ALT_ID          |     1 |    19 |     2   (0)| 00:00:01 |
|* 19 |         INDEX RANGE SCAN                  | ODS_PRODUCT_ALT_ID_PK       |     1 |       |     1   (0)| 00:00:01 |
|* 20 |       TABLE ACCESS BY INDEX ROWID    | ODS_CUST_ACCT_ALT_ID        |     1 |    17 |     2   (0)| 00:00:01 |
|* 21 |        INDEX RANGE SCAN                   | ODS_CUST_ACCT_ALT_ID_PK     |     1 |       |     1   (0)| 00:00:01 |
| 22 |      TABLE ACCESS BY INDEX ROWID     | ODS_TRANS_EXTNN             |     1 |    14 |     3   (0)| 00:00:01 |
|* 23 |       INDEX UNIQUE SCAN                   | SYS_C0026927                |     1 |       |     2   (0)| 00:00:01 |
|* 24 |     INDEX RANGE SCAN                      | ODS_TRANS_ITEM_PK           |     1 |       |     3   (0)| 00:00:01 |
--------------------------------------------------------------------------------------------------------------------Cant avoid the order by and distinct clauses. Cant avoid Outer joins. Indexes are used properly.

Just to add on to Mark's excellent points--
1) The Oracle optimizer is estimating that your query is only going to return 79 rows. If it is using a great deal of TEMP space, that estimate would seem to be wildly incorrect which implies that you have a problem with your statistics.
2) Are you certain that you need the DISTINCT? Lots of people either throw DISTINCT clauses into their queries "just in case" or throw in a distinct when they are unexpectedly getting duplicate rows because they are missing a join condition. If that's the case here, removing the DISTINCT and figuring out which join condition is missing (if applicable) would cause much less TEMP space to be used.
Justin

Similar Messages

When using large text my lines overpal, how can I increase the line spacing?

When using large text my lines overpal, how can I increase the line spacing?

Press 'Ctrl + Enter' after the numbered item to insert blank space below it. If you press 'Enter', the next line will start with the next number. As long as you want to type lines without a numbering, press 'Ctrl + Enter' and type the line.

CLOB processing consuming large memory in temp tablespace

We are having a stored procedure, which is called from a Java process to enqeue a message in to AQ. The input for the procedure is CLOB.
In some installations where oracle standard editions is used, the temp tablespace is growing in large volumes and the only process accessing the temp tablespace is the procedure. Oracle version is 9i 9.2.0.5.0
We are doing a trim on the clob variable inside the procedure. Will that cause any issues like this ?
Procedure is as below
procedure enq_msg
(queue_name varchar2, msg_text in clob,p_result out number) as
queue_options DBMS_AQ.ENQUEUE_OPTIONS_T;
message_properties DBMS_AQ.MESSAGE_PROPERTIES_T;
message_id RAW(16);
v_msg_text               clob default msg_text;
v_queue_name          varchar2(40) deafult queue_name;
BEGIN
v_msg_text := trim(v_msg_text);
v_queue_name := queue_name;
DBMS_AQ.ENQUEUE(queue_name => v_queue_name,
                         enqueue_options => queue_options,
                         message_properties => message_properties,
                         payload => v_msg_text,
                         msgid => message_id
     commit;
     p_result := 0;
exception
when no_data_found then
p_result := 1;
     rollback;
when others then
     rollback;
p_result := sqlcode;
END enq_msg;

Is there a reason you have v_msg_text and v_queue_name defined? You can use queue_name directly in the call to dbms_aq.enqueue and also payload => trim(msg_text)
This would eliminate the copying of the clob that is currently occurring and may help your space issue.
Is trim really needed?
I'm not familiar with any CLOB related bugs in 9i but the above is just some things I would try or question.

How to get query used in LOV

Hi all,,,
Assume that my lov is listing (eg empno, ename,dept) when i query for empno. Actually i want to change that empno to deptno ie(deptno,ename,dept) when i queried for empno. How to accomplish this. I think by form personalization we can do ie. BY creating own record group and replacing that with old one, we can do. But i want to know the query used in old record group. How to accomplish this task. plz help me.. Its urgent ... plz... Advance thanks. Any other approach most welcome.. Plz

Hi all,,,
Assume that my lov is listing (eg empno, ename,dept) when i query for empno. Actually i want to change that empno to deptno ie(deptno,ename,dept) when i queried for empno. How to accomplish this. I think by form personalization we can do ie. BY creating own record group and replacing that with old one, we can do. But i want to know the query used in old record group. How to accomplish this task. plz help me.. Its urgent ... plz... Advance thanks. Any other approach most welcome.. Plz

To shrink the size of TEMP tablespace

Dear all,
There is a databse with RAC, now in OEM the size of TEMP tablespace has been reached at 99.9%. now we want to shrink the size of TEMP tablespace.
how to we do that???????
plz help me...........

Temporary tablespaces usually show they are full, however this space is not actually in use. It is rather allocated. Oracle has evaluated the best way to obtain the most of performance, and he said it is better to allocate once than allocate-deallocate-reallocate extents, so temporary space is not 'released'.
If you want to feel psychologically more confortable with lower allocated space, you can drop your tablespace (create an interim default temporary tablespace first) and recreate it.
You can also rebuild temporary datafiles:
alter tablespace temp add tempfile 'C:\ORACLE\ORACLEXE\ORADATA\XE\TEMP01.dbf' size 32m;
SQL> select name from v$tempfile;
NAME
C:\ORACLE\ORACLEXE\ORADATA\XE\TEMP.DBF
C:\ORACLE\ORACLEXE\ORADATA\XE\TEMP01.DBF
SQL> alter database tempfile 'C:\ORACLE\ORACLEXE\ORADATA\XE\TEMP01.DBF' drop including datafiles;
Database altered.

CMS Repository Database Temp Tablespace Sizing

1. While running some large reports Business Objects (XI 3.1 FP2.7) received an ORA-01652 (Unable to extend temp segment in tablespace TEMP) error for the CMS repository. The CMS database is running in an independent Oracle instance with the auditing database, so only the Business Objects engine would be using the temp tablespace. All the report information is extracted from other database instances. This tablespace was sized at 4GB when this error occurred.
2. What is Business Objects using the CMS repository temp tablespace for? What are some sizing guidelines, should this tablespace be sized based on estimated report sizes, report rows, concurrent users, etc?

I'd say this has nothing to do with Webi processing and has everything to do with your Auditing DB.
Auding DB always growing and that's where that message comes.
Audit guide has more info on managing Audit DB.

ORA-01652 in TEMP Tablespace

Hi,
We have the following errors:
RMAN> crosscheck archivelog all;
starting full resync of recovery catalog
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of crosscheck command at 01/07/2009 14:26:10
RMAN-03014: implicit resync of recovery catalog failed
RMAN-03009: failure of full resync command on default channel at 01/07/2009 14:26:10
ORA-01652: unable to extend temp segment by in tablespace
RMAN>
We have tried to increase the size of the TEMP tablespace, but this operation keeps using up the entire temp tablespace. It's the temp tablespace in the target database that is filling up.
We cannot keep increasing the size of the TEMP ts to fill the disk.
We have also tried to create a new repository but this had the same outcome.
I reckon there is something in the database tables, as one of our backups used the target control file as its repository. Should I remove the entries from these tables in the target tablespace to make the repository think there is no re-synching to be done?
Regards,
Tim.

If you are sure that it is on the target DB what you can do is enable a trace for the event to see what triggers this error
it doesnt have to be on temporary table because there are temp segments on normal tables used for index creation type operations as well so you need to find the problem before assuming it is temp tablespace error
alter system set events '1652 trace name errorstack level 1';
to turn off
alter system set events '1652 trace name context off';
If you cant find anything from trace then you can send the trace output here maybe somebody can catch something.
Coskan Gundogar
http://coskan.wordpress.com
Edited by: coskan on Mar 30, 2009 4:28 PM

How do i know if my query using SORT_AREA_SIZE   or temporary tablespace ?

Good Morning Everyone !
My DB version is 10.2.0.1
I have large table exactly 3 million records.
SQL> select count(*) from tab1;
COUNT(*)
   300000
SQL> select * from tab1 order by no DESC;
sorting ... in process
300000 rows selected.
in Terminal 2 : I tried to find sorting details -   ( No rows selected - why ? )
SQL> select USERNAME , USER , TABLESPACE , SQL_ID from v$tempseg_usage ;
no rows selected
SQL> /
no rows selected
When i google i have seen this ;
If Oracle cannot do the sort in memory (SORT_AREA_SIZE initialisation parameter), space will be allocated in a temporary tablespace for doing the sort operation.
REF_LINK : TEMPORARY Tablespaces and TEMPFILES | Oracle FAQ
MY DOUBT QUESTION :   How do i know if my query using SORT_AREA_SIZE or temporary tablespace ?
Thanks in advance.

@ JohnWatson
I have seen some articles from ORA - FAQ. Good.
SQL> select USERNAME , USER , TABLESPACE , SQL_ID from v$tempseg_usage;
USERNAME                       USER   TABLESPACE                      SQL_ID
   SCOTT                               SYS          TEMP                            fh9vqgyd6m0d1
PGA management means that sorting only 300000 rows may well occur in memory
            Is this (3 million rows) - standard value for 10g version ?
Thanks JohnWatson

How to exclude UNDO and temp tablespace using monitoring template

hi,
as per MOS note id 816920.1 undo and temp tablespace is excluded from monitoring starting version 11G onwards.It says that thresholds will need to be explicitly set if undo and temp needs to be monitored
we have grid 12C implemented now with monitoring templates which sets thresholds for all tablespaces. does it apply that thresholds to undo and temp too?? we are receiving alerts for undo and temp tablespaces and we wish to disable them. We have undo tablespaces with different names in many databases as per naming conventions for that line of business.
How can i exclude undo and temp tablespaces from monitoring using grid 12C monitoring templates. is there a way through templates where i can set thresholds of permanent tablespaces only???

In EM12c, go to the 'Enterprise' menu, then 'Job', then 'Library'. There should be an out-of-the-box job called "DISABLE TABLESPACE USED (%) ALERTS FOR UNDO AND TEMP TABLESPACES" that you can run against your database targets, and that job will disable database-generated alerts.
If a previous admin has set up EM12c-generated alerts for undo and temp, then yes, a monitoring template will take care of disabling them. Either remove those alerts from the existing monitoring template and apply them to your targets, or create a new monitoring template based on one database's current settings, remove those warning/critical thresholds for your undo and temp tablespaces, and apply it to your targets.

Query on V$STREAMS_APPLY_COORDINATOR used large temp space

Hi
My database recently experience ORA-1652 error consistently on temp tablespace. I found out the query to cause this was an OEM query SELECT APPLY_NAME,TOTAL_RECEIVED,TOTAL_ASSIGNED,TOTAL_APPLIED FROM V$STREAMS_APPLY_COORDINATOR, which uses over than 800M temp space frequently. I am not sure why is this, I can query the same select statement in sqlplus, it returns immeidatly. Can anyone help to teach me on this? Thanks.
The query to find the user for temp tablespace is.
SELECT S.sid || ',' || S.serial# sid_serial, S.username, Q.hash_value, Q.sql_text,
T.blocks * TBS.block_size / 1024 / 1024 mb_used, T.tablespace
FROM v\$sort_usage T, v\$session S, v\$sqlarea Q, dba_tablespaces TBS
WHERE T.session_addr = S.saddr
AND T.sqladdr = Q.address
AND T.tablespace = TBS.tablespace_name
ORDER BY mb_used)
where mb_used >=800;
I got
SID_SERIAL USERNAME HASH_VALUE SQL_TEXT MB_USED TABLESPACE
1626,1 DBSNMP 3741683138 /* OracleOEM */ SELECT APPLY_NAME,TOTAL_RECEIVED,TOTAL_ASSIGNED,TOTAL_APPLIED FROM V$STREAMS_APPLY_COORDINATOR 880 TEMP1

Hi
My database recently experience ORA-1652 error consistently on temp tablespace. I found out the query to cause this was an OEM query SELECT APPLY_NAME,TOTAL_RECEIVED,TOTAL_ASSIGNED,TOTAL_APPLIED FROM V$STREAMS_APPLY_COORDINATOR, which uses over than 800M temp space frequently. I am not sure why is this, I can query the same select statement in sqlplus, it returns immeidatly. Can anyone help to teach me on this? Thanks.
The query to find the user for temp tablespace is.
SELECT S.sid || ',' || S.serial# sid_serial, S.username, Q.hash_value, Q.sql_text,
T.blocks * TBS.block_size / 1024 / 1024 mb_used, T.tablespace
FROM v\$sort_usage T, v\$session S, v\$sqlarea Q, dba_tablespaces TBS
WHERE T.session_addr = S.saddr
AND T.sqladdr = Q.address
AND T.tablespace = TBS.tablespace_name
ORDER BY mb_used)
where mb_used >=800;
I got
SID_SERIAL USERNAME HASH_VALUE SQL_TEXT MB_USED TABLESPACE
1626,1 DBSNMP 3741683138 /* OracleOEM */ SELECT APPLY_NAME,TOTAL_RECEIVED,TOTAL_ASSIGNED,TOTAL_APPLIED FROM V$STREAMS_APPLY_COORDINATOR 880 TEMP1

How to find the sessions/user consuming more temp tablespace

Environment details
IBM-AIX 64 bit
Oracle 10.2.0.4.0
Recently we encountered the issue of sudden peak in CPU utilization and temp tablespace usage. We are noticing this issue after Apr CPU 2009 patch and upgrade to 10.2.0.4.
Recently temp space was full and we added 18 GB of space to temp tablespace, within few hours, they are also consumed.
Application team is telling that they did not modify any code. ADDM report suggests most of the recommendations related to SQL tuning only
not sure about the real issue. Can somebody tell me how I can find what are all sessions/user consuming high temp tablespace
Edited by: user1368801 on Sep 1, 2009 5:32 PM

Hi,
Run below query to check if it is used by any session.
select sum(a.BYTES_USED)/1024/1024 used_mb, sum(b.bytes)/1024/1024 total_mb,
sum(a.BYTES_USED)/sum(b.bytes)*100 pct_used
from V$TEMP_EXTENT_POOL a, v$tempfile b
where a.file_id(+) = b.file#
Check who is using it by the below query.
SELECT s.sid,s.serial#,osuser,process,program,s.sql_hash_value,u.extents, u.blocks
FROM v\$session s, v\$sort_usage u
WHERE s.saddr = u.session_addr order by extents,blocks desc
Hope this solves your issue.
Regards
Regards,
Satishbabu Gunukula
Click here to improve RMAN backup Performance using [Block change tracking in Oracle 10g|http://oracleracexpert.blogspot.com/2009/09/block-change-tracking-in-oracle-10g.html]
http://oracleracexpert.blogspot.com

How to create temp files in temp tablespace

Dear all,
Due to outage of our SAN, we our out of production for the
last 3 days. By the grace of Almighty we have restored production by
database recovery from our standby backup. Since temp tablespace and
temp data files do not taken as backup for standby, now after recovery
we are getting abap dumps asking for temp_1 and temp_2 datafiles.
Please guide us how to create temp files. v$tablespace is showing tablespace PSAPTEMP but datafile are not there
Abap dumps are giving these errors as mentioned below
====================================================
The exception must either be prevented, caught within the procedure
"DATA_SELECTION"
"(FORM)", or declared in the procedure's RAISING clause.
To prevent the exception, note the following:
Database error text........: "ORA-01157: cannot identify/lock data file 256 -
see DBWR trace file#ORA-01110: data file 256:
'/oracle/SD1/sapdata4/temp_2/temp.data2'"
Internal call code.........: "[RSQL/FTCH/MARA ]"
===================================================
another one asking for
'/oracle/SD1/sapdata3/temp_1/temp.data1'"
Best Regards
Waqas

if you want to add a new tempfile to your TEMP Tablespace,you can do like that.
ALTER TABLESPACE <name of TEMP Tablespace> ADD TEMPFILE <pfad to the file_and_file name> SIZE <size>;
You can use also the options REUSE autoextend off or on .
e.g:
alter tablespace PSAPTEMP add tempfile '/oracle/SD1/sapdata4/temp_2/temp.data2' SIZE 1000K REUSE;
The directory temp_2 should exist.
Or you can use the BR*Tools to create a new datafile. Enter brtools and follow the instructions or menu.
More to TEMP Tablespaces see please following SAP notes:
659946 - FAQ: Temporary tablespaces
600513 - ORA-25153 after recovery due to missing tempfiles
and the Oracle Note:
160426.1: TEMPORARY Tablespaces : Tempfiles or Datafiles ?
I hope it helps.

Oracle 9i TEMP tablespace backup problem using RMAN!

Oracle8/8i whole backup is ok for our backup software(using RMAN without RC database), but for Oracle 9i, I get following error messages when backing up temp tablespace.
1. RMAN-20202: tablespace not found in the recovery catalog
2. RMAN-06019: could not translate tablespace name "TEMP"
I check some views of Oracle9i, and know some changes happen for temp tablespace in 9i, but how to deal with this problem. Any idea, please!

In 9i RMAN does not restore temporary datafiles. Instead, you should create them after your restore. I believe that Oracle is going to make a change to this in the next release of 9i and have RMAN create the temporary files. You can view the temporary datafiles @ v$tempfile.
I believe RMAN doesn't restore temporary files because they are locally managed and not in the control files. RMAN reads the controlfile of the target database to obtain info about backups, datafiles, etc.
Hope this helps.

How do I...Mimic Replacement Path with Query using Customer Variable Exits?

Hello SDN:
We are on BW 3.5 SP16.
We are currently using a replacement path Query(RP) with variables
to populate a variable in another Query(T).
The problem we are having is with performance.
The entry of variables in replacement path Query(RP) is optional (This is necessary there cannot be required values)
When results from Replacement Path Query are small performance is fine. (e.g. 10 seconds)
When results from Replacement path query are large performance suffers. (e.g. 1+ minutes)
Users are free to leave the replacement path variables empty resulting in a large set of data to be replaced. This is the worst performance case.
We would like to discover a way to conditionally execute the replacement path query. That is if users do not enter values for the replacement Query(RP) variables do not execute the replacement path query(RP).
Does anyone know if this is possible within reason and in customer exit space?
We have reviewed the situation from all angles and the requirement for the replacement path FUNCTIONALITY and the freedom for the user to leave variable values blank remains.
I've been searching and reading SDN and SAP notes for about a week and do not find threads which address this situation.
We are also exploring Customer variable exits to mimic replacement path functionality
(different topic subject="How do I...Mimic Replacement Path with Query using Customer Variable Exits?")
Any help will be appreciated
Many thanks
David Schuh

My appologies-I posted this message with the wrong subject. I will repost it with appropriate subject.
dave schuh

How to monitor temp tablespace

I server in 11gR2 and 10gR2.
Is there a way to monitor its usage, like, who is using it, how much space a session is using, what are the objects crated, ie. temp tables.
Thanks.

Hi
This might helpful
select TABLESPACE_NAME, BYTES_USED, BYTES_FREE from V$TEMP_SPACE_HEADER;Also checkout the below link.
http://blog.flimatech.com/2010/06/15/how-to-monitor-temp-tablespace-usage/
Cheers
Kanchana

Query using large(in GB) temp tablespace, how to tune it?

Similar Messages

Maybe you are looking for